Hugging Face's logo

FlorianJc
/
MegaBeam-Mistral-7B-300k-vllm-fp8

Text Generation
Transformers
Safetensors
mistral
vllm
fp8
conversational
text-generation-inference
Model card Files
xet
Community
FlorianJc commited on
Commit
2a3f386
·
verified ·
1 Parent(s): e06b053

Init

Browse files
Files changed (10) hide show
  1. README.md +410 -0
  2. config.json +33 -0
  3. generation_config.json +6 -0
  4. model-00001-of-00002.safetensors +3 -0
  5. model-00002-of-00002.safetensors +3 -0
  6. model.safetensors.index.json +746 -0
  7. special_tokens_map.json +23 -0
  8. tokenizer.json +0 -0
  9. tokenizer.model +3 -0
  10. tokenizer_config.json +44 -0
README.md ADDED
@@ -0,0 +1,410 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ quantized_by: FlorianJc
3
+ license: apache-2.0
4
+ inference: false
5
+ ---
6
+
7
+
8
+ ## Model infos:
9
+ [MegaBeam-Mistral-7B-300k](https://huggingface.co/amazon/MegaBeam-Mistral-7B-300k) quantized to FP8 weights and activations using per-tensor quantization, ready for inference with vLLM >= 0.5.1.
10
+
11
+
12
+ # Original model README.md file:
13
+
14
+
15
+ # MegaBeam-Mistral-7B-300k Model
16
+
17
+ MegaBeam-Mistral-7B-300k is a fine-tuned [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) language model that supports input contexts up to 320k tokens. MegaBeam-Mistral-7B-300k can be deployed on a single AWS `g5.48xlarge` instance using serving frameworks such as [vLLM](https://github.com/vllm-project/vllm), Sagemaker [DJL](https://docs.aws.amazon.com/sagemaker/latest/dg/deploy-models-frameworks-djl-serving.html) endpoint, and others. Similarities and differences beween MegaBeam-Mistral-7B-300k and [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) are summarized below:
18
+
19
+
20
+ |Model|Max context length| rope_theta| prompt template|
21
+ |----------|-------------:|------------:|------------:|
22
+ | [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) | 32K | 1e6 | [instruction format](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2#instruction-format)|
23
+ | MegaBeam-Mistral-7B-300k | 320K | 25e6 | AS ABOVE|
24
+
25
+ ## Evaluations
26
+
27
+ **[InfiniteBench: Extending Long Context Evaluation Beyond 100K Tokens](https://github.com/OpenBMB/InfiniteBench)**
28
+
29
+ _InfiniteBench is a cutting-edge benchmark tailored for evaluating the capabilities of language models to process, understand, and reason over super long contexts (100k+ tokens)_. We therefore evaluated MegaBeam-Mistral-7B-300k, [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2), [Llama-3-8B-Instruct-262k](https://huggingface.co/gradientai/Llama-3-8B-Instruct-262k), and [Llama3-70B-1M](https://huggingface.co/gradientai/Llama-3-70B-Instruct-Gradient-1048k) on InfiniteBench. The InfiniteBench authors also evaluated SOTA proprietary and open-source LLMs on InfiniteBench. We thus combined both results in the table below.
30
+
31
+ | Task Name | MegaBeam-Mistral-7B-300k | Mistral-7B-Instruct-v0.2 | Llama-3-8B-Instruct-262k | Llama3-70B-1M | GPT-4-1106-preview | YaRN-Mistral-7B | Kimi-Chat | Claude 2 | Yi-6B-200K | Yi-34B-200K | Chatglm3-6B-128K |
32
+ | ---------------- | ---------------- | ---------------- | ---------------- | ---------------- | ------ | --------------- | --------- | -------- | -----------| -----------| -----------|
33
+ | Retrieve.PassKey | 100% | 75.76% | 98.30% | 81.35% | 100% | 92.71% | 98.14% | 97.80% | 100.00% | 100.00% | 92.20% |
34
+ | Retrieve.Number | 96.10% | 25.25% | 97.79% | 97.62% | 100% | 56.61% | 95.42% | 98.14% | 94.92% | 100.00% | 80.68% |
35
+ | Retrieve.KV | 0% | 0% | 3.40% | 3% | 89.00% | < 5% | 53.60% | 65.40% | < 5% | < 5% | < 5% |
36
+ | En.Sum | 29.39% | 22.13% | 16.40% | 20.72% | 14.73% | 9.09% | 17.93% | 14.45% | < 5% | < 5% |< 5% |
37
+ | En.QA | 14.93% | 4.93% | 13.20% | 16.52% | 22.22% | 9.55% | 16.52% | 11.97% | 9.20% | 12.17% |< 5% |
38
+ | En.MC | 51.52% | 7.80% | 50.65% | 62% | 67.25% | 27.95% | 72.49% | 62.88% | 36.68% |38.43% |10.48% |
39
+ | En.Dia | 9.50% | 3.50% | 1% | 12.50% | 8.50% | 7.50% | 11.50% | 46.50% | < 5% |< 5% |< 5% |
40
+ | Zh.QA | 10.71% | 3.43% | 19.02% | 26% | 25.96% | 14.43% | 17.93% | 9.64% | 15.07% |13.61% |< 5% |
41
+ | Code.Debug | 27.41% | 11.60% | 22.08% | 23.85% | 39.59% | < 5% | 18.02% | < 5% | < 5% |< 5% |< 5% |
42
+ | Code.Run | 1.75% | 0.25% | 0% | 0% | 23.25% | < 5% | < 5% | < 5% | < 5% |< 5% |< 5% |
43
+ | Math.Calc | 0% | 0% | 0% | 0% | < 5% | < 5% | < 5% | < 5% | < 5% |< 5% |< 5% |
44
+ | Math.Find | 24.28% | 26.28% | 15.40% | 30% | 60.00% | 17.14% | 12.57% | 32.29% | < 5% |25.71% |7.71% |
45
+ | **Average** | 30.70% | 15.08% | 28.10% | 31.13% | 46.08% | 20.41% | 34.93% | 37.21% | 22.78% |25.41% |17.59% |
46
+
47
+ The 12 evaluation tasks are summarized below (as per [InfiniteBench]((https://github.com/OpenBMB/InfiniteBench)))
48
+ | Task Name | Context | # Examples | Avg Input Tokens | Avg Output Tokens | Description |
49
+ | -------------------- | ------------- | ---------- | ---------------- | ----------------- | ------------------------------------------------------------------------------------------- |
50
+ | En.Sum | Fake Book | 103 | 171.5k | 1.1k | Summarization of a fake book created with core entity substitution. |
51
+ | En.QA | Fake Book | 351 | 192.6k | 4.8 | Free-form question answering based on the fake book. |
52
+ | En.MC | Fake Book | 229 | 184.4k | 5.3 | Multiple choice questions derived from the fake book. |
53
+ | En.Dia | Script | 200 | 103.6k | 3.4 | Identification of talkers in partially anonymized scripts. |
54
+ | Zh.QA | New Book | 175 | 2068.6k | 6.3 | Question answering on a set of newly collected books. |
55
+ | Code.Debug | Code Document | 394 | 114.7k | 4.8 | Finding which function in a code repo contains an crashing error (in multiple choice form). |
56
+ | Code.Run | Synthetic | 400 | 75.2k | 1.3 | Simulating execution of multiple simple, synthetic functions. |
57
+ | Math.Calc | Synthetic | 50 | 43.9k | 43.9k | Calculations involving super-long arithmetic equations. |
58
+ | Math.Find | Synthetic | 350 | 87.9k | 1.3 | Finding special integers in a lengthy list. |
59
+ | Retrieve.PassKey | Synthetic | 590 | 122.4k | 2.0 | Retrieving hidden keys in a noisy long context. |
60
+ | Retrieve.Number | Synthetic | 590 | 122.4k | 4.0 | Locating repeated hidden numbers in a noisy long context. |
61
+ | Retrieve.KV | Synthetic | 500 | 89.9k | 22.7 | Finding the corresponding value from a dictionary and a key. |
62
+
63
+
64
+ ## Serve MegaBeam-Mistral-7B-300k on EC2 instances ##
65
+ On an AWS `g5.48xlarge` instance, upgrade vLLM to the latest version as per [documentation on vLLM](https://vllm.readthedocs.io/en/latest/).
66
+
67
+ ### Start the server
68
+ ```shell
69
+ python3 -m vllm.entrypoints.openai.api_server --model amazon/MegaBeam-Mistral-7B-300k --tensor-parallel-size 8
70
+ ```
71
+ **Important Note** - We have set the `max_position_embeddings` in the [`config.json`](config.json) to 288,800 in order to fit model's KV-cache on a single `g5.48xlarge` instance, which has 8 x A10 GPUs (24GB RAM per GPU).
72
+
73
+ On an instance with larger GPU RAM (e.g. `p4d.24xlarge`), feel free to increase the value of the `max_position_embeddings`(e.g. to 350K), which the model should be able to process.
74
+
75
+ ### Run the client
76
+ ```python
77
+ from openai import OpenAI
78
+
79
+ # Modify OpenAI's API key and API base to use vLLM's API server.
80
+ openai_api_key = "EMPTY"
81
+ openai_api_base = "http://localhost:8000/v1"
82
+
83
+ client = OpenAI(
84
+ # defaults to os.environ.get("OPENAI_API_KEY")
85
+ api_key=openai_api_key,
86
+ base_url=openai_api_base,
87
+ )
88
+
89
+ models = client.models.list()
90
+ model = models.data[0].id
91
+
92
+ chat_completion = client.chat.completions.create(
93
+ messages = [
94
+ {"role": "user", "content": "What is your favourite condiment?"}, # insert your long context here
95
+ {"role": "assistant", "content": "Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!"},
96
+ {"role": "user", "content": "Do you have mayonnaise recipes?"} # insert your long context here
97
+ ],
98
+ model=model,
99
+ )
100
+
101
+ print("Chat completion results:")
102
+ print(chat_completion)
103
+ ```
104
+
105
+ ### Deploy the model on a SageMaker Endpoint ###
106
+ To deploy MegaBeam-Mistral-7B-300k on a SageMaker endpoint, please follow this [SageMaker DJL deployment guide](https://docs.djl.ai/docs/demos/aws/sagemaker/large-model-inference/sample-llm/vllm_deploy_mistral_7b.html).
107
+
108
+ Run the following Python code in a SageMaker notebook (with each block running in a separate cell)
109
+
110
+ ```python
111
+ import sagemaker
112
+ from sagemaker import Model, image_uris, serializers, deserializers
113
+
114
+ sagemaker_session = sagemaker.Session()
115
+ region = sagemaker_session.boto_region_name
116
+ role = sagemaker.get_execution_role()
117
+
118
+ %%writefile serving.properties
119
+ engine=Python
120
+ option.model_id=amazon/MegaBeam-Mistral-7B-300k
121
+ option.dtype=bf16
122
+ option.task=text-generation
123
+ option.rolling_batch=vllm
124
+ option.tensor_parallel_degree=8
125
+ option.device_map=auto
126
+
127
+ %%sh
128
+ mkdir mymodel
129
+ mv serving.properties mymodel/
130
+ tar czvf mymodel.tar.gz mymodel/
131
+ rm -rf mymodel
132
+
133
+ image_uri = image_uris.retrieve(
134
+ framework="djl-deepspeed",
135
+ region=region,
136
+ version="0.27.0"
137
+ )
138
+
139
+ s3_code_prefix = "megaBeam-mistral-7b-300k/code"
140
+ bucket = sagemaker_session.default_bucket() # bucket to house artifacts
141
+ code_artifact = sagemaker_session.upload_data("mymodel.tar.gz", bucket, s3_code_prefix)
142
+ print(f"S3 Code or Model tar ball uploaded to --- &gt; {code_artifact}")
143
+ model = Model(image_uri=image_uri, model_data=code_artifact, role=role)
144
+
145
+ instance_type = "ml.g5.48xlarge"
146
+ endpoint_name = sagemaker.utils.name_from_base("megaBeam-mistral-7b-300k")
147
+ model.deploy(initial_instance_count=1,
148
+ instance_type=instance_type,
149
+ endpoint_name=endpoint_name
150
+ )
151
+
152
+ # our requests and responses will be in json format so we specify the serializer and the deserializer
153
+ predictor = sagemaker.Predictor(
154
+ endpoint_name=endpoint_name,
155
+ sagemaker_session=sagemaker_session,
156
+ serializer=serializers.JSONSerializer(),
157
+ )
158
+
159
+ # test the endpoint
160
+ input_str = """<s>[INST] What is your favourite condiment? [/INST]
161
+ Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!</s> "
162
+ [INST] Do you have mayonnaise recipes? [/INST]"""
163
+ predictor.predict(
164
+ {"inputs": input_str, "parameters": {"max_new_tokens": 75}}
165
+ )
166
+
167
+ ```
168
+
169
+ ### Invoke the model on a SageMaker Endpoint ###
170
+ To use MegaBeam-Mistral-7B-300k on a SageMaker endpoint, please try following this example:
171
+
172
+ ```python
173
+ import boto3
174
+ import json
175
+
176
+ def call_endpoint(text:str, endpoint_name:str):
177
+ client = boto3.client("sagemaker-runtime")
178
+
179
+ parameters = {
180
+ "max_new_tokens": 450,
181
+ "do_sample": True,
182
+ "temperature": 0.7,
183
+ }
184
+
185
+ payload = {"inputs": text, "parameters": parameters}
186
+
187
+ response = client.invoke_endpoint(
188
+ EndpointName=endpoint_name, Body=json.dumps(payload), ContentType="application/json"
189
+ )
190
+
191
+ output = json.loads(response["Body"].read().decode())
192
+
193
+ result = output["generated_text"]
194
+ return result
195
+
196
+ # please insert your long prompt/document content here
197
+ prompt = """<s>[INST] What are the main challenges to support long contexts for a Large Language Model? [/INST]"""
198
+
199
+ #print(prompt)
200
+ endpoint_name = "megaBeam-mistral-7b-300k-2024-05-13-14-23-41-219" # please use a valid endpoint name
201
+ result = call_endpoint(prompt, endpoint_name)
202
+ print(result)
203
+ ```
204
+
205
+
206
+ ## Limitations ##
207
+ Before using the MegaBeam-Mistral-7B-300k model, it is important to perform your own independent assessment, and take measures to ensure that your use would comply with your own specific quality control practices and standards, and that your use would comply with the local rules, laws, regulations, licenses and terms that apply to you, and your content.
208
+
209
+ ## The AWS Contributors ##
210
+ Chen Wu, Yin Song, Verdi March, Eden Duthi---
211
+ license: apache-2.0
212
+ inference: false
213
+ ---
214
+
215
+ # MegaBeam-Mistral-7B-300k Model
216
+
217
+ MegaBeam-Mistral-7B-300k is a fine-tuned [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) language model that supports input contexts up to 320k tokens. MegaBeam-Mistral-7B-300k can be deployed on a single AWS `g5.48xlarge` instance using serving frameworks such as [vLLM](https://github.com/vllm-project/vllm), Sagemaker [DJL](https://docs.aws.amazon.com/sagemaker/latest/dg/deploy-models-frameworks-djl-serving.html) endpoint, and others. Similarities and differences beween MegaBeam-Mistral-7B-300k and [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) are summarized below:
218
+
219
+
220
+ |Model|Max context length| rope_theta| prompt template|
221
+ |----------|-------------:|------------:|------------:|
222
+ | [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) | 32K | 1e6 | [instruction format](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2#instruction-format)|
223
+ | MegaBeam-Mistral-7B-300k | 320K | 25e6 | AS ABOVE|
224
+
225
+ ## Evaluations
226
+
227
+ **[InfiniteBench: Extending Long Context Evaluation Beyond 100K Tokens](https://github.com/OpenBMB/InfiniteBench)**
228
+
229
+ _InfiniteBench is a cutting-edge benchmark tailored for evaluating the capabilities of language models to process, understand, and reason over super long contexts (100k+ tokens)_. We therefore evaluated MegaBeam-Mistral-7B-300k, [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2), [Llama-3-8B-Instruct-262k](https://huggingface.co/gradientai/Llama-3-8B-Instruct-262k), and [Llama3-70B-1M](https://huggingface.co/gradientai/Llama-3-70B-Instruct-Gradient-1048k) on InfiniteBench. The InfiniteBench authors also evaluated SOTA proprietary and open-source LLMs on InfiniteBench. We thus combined both results in the table below.
230
+
231
+ | Task Name | MegaBeam-Mistral-7B-300k | Mistral-7B-Instruct-v0.2 | Llama-3-8B-Instruct-262k | Llama3-70B-1M | GPT-4-1106-preview | YaRN-Mistral-7B | Kimi-Chat | Claude 2 | Yi-6B-200K | Yi-34B-200K | Chatglm3-6B-128K |
232
+ | ---------------- | ---------------- | ---------------- | ---------------- | ---------------- | ------ | --------------- | --------- | -------- | -----------| -----------| -----------|
233
+ | Retrieve.PassKey | 100% | 75.76% | 98.30% | 81.35% | 100% | 92.71% | 98.14% | 97.80% | 100.00% | 100.00% | 92.20% |
234
+ | Retrieve.Number | 96.10% | 25.25% | 97.79% | 97.62% | 100% | 56.61% | 95.42% | 98.14% | 94.92% | 100.00% | 80.68% |
235
+ | Retrieve.KV | 0% | 0% | 3.40% | 3% | 89.00% | < 5% | 53.60% | 65.40% | < 5% | < 5% | < 5% |
236
+ | En.Sum | 29.39% | 22.13% | 16.40% | 20.72% | 14.73% | 9.09% | 17.93% | 14.45% | < 5% | < 5% |< 5% |
237
+ | En.QA | 14.93% | 4.93% | 13.20% | 16.52% | 22.22% | 9.55% | 16.52% | 11.97% | 9.20% | 12.17% |< 5% |
238
+ | En.MC | 51.52% | 7.80% | 50.65% | 62% | 67.25% | 27.95% | 72.49% | 62.88% | 36.68% |38.43% |10.48% |
239
+ | En.Dia | 9.50% | 3.50% | 1% | 12.50% | 8.50% | 7.50% | 11.50% | 46.50% | < 5% |< 5% |< 5% |
240
+ | Zh.QA | 10.71% | 3.43% | 19.02% | 26% | 25.96% | 14.43% | 17.93% | 9.64% | 15.07% |13.61% |< 5% |
241
+ | Code.Debug | 27.41% | 11.60% | 22.08% | 23.85% | 39.59% | < 5% | 18.02% | < 5% | < 5% |< 5% |< 5% |
242
+ | Code.Run | 1.75% | 0.25% | 0% | 0% | 23.25% | < 5% | < 5% | < 5% | < 5% |< 5% |< 5% |
243
+ | Math.Calc | 0% | 0% | 0% | 0% | < 5% | < 5% | < 5% | < 5% | < 5% |< 5% |< 5% |
244
+ | Math.Find | 24.28% | 26.28% | 15.40% | 30% | 60.00% | 17.14% | 12.57% | 32.29% | < 5% |25.71% |7.71% |
245
+ | **Average** | 30.70% | 15.08% | 28.10% | 31.13% | 46.08% | 20.41% | 34.93% | 37.21% | 22.78% |25.41% |17.59% |
246
+
247
+ The 12 evaluation tasks are summarized below (as per [InfiniteBench]((https://github.com/OpenBMB/InfiniteBench)))
248
+ | Task Name | Context | # Examples | Avg Input Tokens | Avg Output Tokens | Description |
249
+ | -------------------- | ------------- | ---------- | ---------------- | ----------------- | ------------------------------------------------------------------------------------------- |
250
+ | En.Sum | Fake Book | 103 | 171.5k | 1.1k | Summarization of a fake book created with core entity substitution. |
251
+ | En.QA | Fake Book | 351 | 192.6k | 4.8 | Free-form question answering based on the fake book. |
252
+ | En.MC | Fake Book | 229 | 184.4k | 5.3 | Multiple choice questions derived from the fake book. |
253
+ | En.Dia | Script | 200 | 103.6k | 3.4 | Identification of talkers in partially anonymized scripts. |
254
+ | Zh.QA | New Book | 175 | 2068.6k | 6.3 | Question answering on a set of newly collected books. |
255
+ | Code.Debug | Code Document | 394 | 114.7k | 4.8 | Finding which function in a code repo contains an crashing error (in multiple choice form). |
256
+ | Code.Run | Synthetic | 400 | 75.2k | 1.3 | Simulating execution of multiple simple, synthetic functions. |
257
+ | Math.Calc | Synthetic | 50 | 43.9k | 43.9k | Calculations involving super-long arithmetic equations. |
258
+ | Math.Find | Synthetic | 350 | 87.9k | 1.3 | Finding special integers in a lengthy list. |
259
+ | Retrieve.PassKey | Synthetic | 590 | 122.4k | 2.0 | Retrieving hidden keys in a noisy long context. |
260
+ | Retrieve.Number | Synthetic | 590 | 122.4k | 4.0 | Locating repeated hidden numbers in a noisy long context. |
261
+ | Retrieve.KV | Synthetic | 500 | 89.9k | 22.7 | Finding the corresponding value from a dictionary and a key. |
262
+
263
+
264
+ ## Serve MegaBeam-Mistral-7B-300k on EC2 instances ##
265
+ On an AWS `g5.48xlarge` instance, upgrade vLLM to the latest version as per [documentation on vLLM](https://vllm.readthedocs.io/en/latest/).
266
+
267
+ ### Start the server
268
+ ```shell
269
+ python3 -m vllm.entrypoints.openai.api_server --model amazon/MegaBeam-Mistral-7B-300k --tensor-parallel-size 8
270
+ ```
271
+ **Important Note** - We have set the `max_position_embeddings` in the [`config.json`](config.json) to 288,800 in order to fit model's KV-cache on a single `g5.48xlarge` instance, which has 8 x A10 GPUs (24GB RAM per GPU).
272
+
273
+ On an instance with larger GPU RAM (e.g. `p4d.24xlarge`), feel free to increase the value of the `max_position_embeddings`(e.g. to 350K), which the model should be able to process.
274
+
275
+ ### Run the client
276
+ ```python
277
+ from openai import OpenAI
278
+
279
+ # Modify OpenAI's API key and API base to use vLLM's API server.
280
+ openai_api_key = "EMPTY"
281
+ openai_api_base = "http://localhost:8000/v1"
282
+
283
+ client = OpenAI(
284
+ # defaults to os.environ.get("OPENAI_API_KEY")
285
+ api_key=openai_api_key,
286
+ base_url=openai_api_base,
287
+ )
288
+
289
+ models = client.models.list()
290
+ model = models.data[0].id
291
+
292
+ chat_completion = client.chat.completions.create(
293
+ messages = [
294
+ {"role": "user", "content": "What is your favourite condiment?"}, # insert your long context here
295
+ {"role": "assistant", "content": "Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!"},
296
+ {"role": "user", "content": "Do you have mayonnaise recipes?"} # insert your long context here
297
+ ],
298
+ model=model,
299
+ )
300
+
301
+ print("Chat completion results:")
302
+ print(chat_completion)
303
+ ```
304
+
305
+ ### Deploy the model on a SageMaker Endpoint ###
306
+ To deploy MegaBeam-Mistral-7B-300k on a SageMaker endpoint, please follow this [SageMaker DJL deployment guide](https://docs.djl.ai/docs/demos/aws/sagemaker/large-model-inference/sample-llm/vllm_deploy_mistral_7b.html).
307
+
308
+ Run the following Python code in a SageMaker notebook (with each block running in a separate cell)
309
+
310
+ ```python
311
+ import sagemaker
312
+ from sagemaker import Model, image_uris, serializers, deserializers
313
+
314
+ sagemaker_session = sagemaker.Session()
315
+ region = sagemaker_session.boto_region_name
316
+ role = sagemaker.get_execution_role()
317
+
318
+ %%writefile serving.properties
319
+ engine=Python
320
+ option.model_id=amazon/MegaBeam-Mistral-7B-300k
321
+ option.dtype=bf16
322
+ option.task=text-generation
323
+ option.rolling_batch=vllm
324
+ option.tensor_parallel_degree=8
325
+ option.device_map=auto
326
+
327
+ %%sh
328
+ mkdir mymodel
329
+ mv serving.properties mymodel/
330
+ tar czvf mymodel.tar.gz mymodel/
331
+ rm -rf mymodel
332
+
333
+ image_uri = image_uris.retrieve(
334
+ framework="djl-deepspeed",
335
+ region=region,
336
+ version="0.27.0"
337
+ )
338
+
339
+ s3_code_prefix = "megaBeam-mistral-7b-300k/code"
340
+ bucket = sagemaker_session.default_bucket() # bucket to house artifacts
341
+ code_artifact = sagemaker_session.upload_data("mymodel.tar.gz", bucket, s3_code_prefix)
342
+ print(f"S3 Code or Model tar ball uploaded to --- &gt; {code_artifact}")
343
+ model = Model(image_uri=image_uri, model_data=code_artifact, role=role)
344
+
345
+ instance_type = "ml.g5.48xlarge"
346
+ endpoint_name = sagemaker.utils.name_from_base("megaBeam-mistral-7b-300k")
347
+ model.deploy(initial_instance_count=1,
348
+ instance_type=instance_type,
349
+ endpoint_name=endpoint_name
350
+ )
351
+
352
+ # our requests and responses will be in json format so we specify the serializer and the deserializer
353
+ predictor = sagemaker.Predictor(
354
+ endpoint_name=endpoint_name,
355
+ sagemaker_session=sagemaker_session,
356
+ serializer=serializers.JSONSerializer(),
357
+ )
358
+
359
+ # test the endpoint
360
+ input_str = """<s>[INST] What is your favourite condiment? [/INST]
361
+ Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!</s> "
362
+ [INST] Do you have mayonnaise recipes? [/INST]"""
363
+ predictor.predict(
364
+ {"inputs": input_str, "parameters": {"max_new_tokens": 75}}
365
+ )
366
+
367
+ ```
368
+
369
+ ### Invoke the model on a SageMaker Endpoint ###
370
+ To use MegaBeam-Mistral-7B-300k on a SageMaker endpoint, please try following this example:
371
+
372
+ ```python
373
+ import boto3
374
+ import json
375
+
376
+ def call_endpoint(text:str, endpoint_name:str):
377
+ client = boto3.client("sagemaker-runtime")
378
+
379
+ parameters = {
380
+ "max_new_tokens": 450,
381
+ "do_sample": True,
382
+ "temperature": 0.7,
383
+ }
384
+
385
+ payload = {"inputs": text, "parameters": parameters}
386
+
387
+ response = client.invoke_endpoint(
388
+ EndpointName=endpoint_name, Body=json.dumps(payload), ContentType="application/json"
389
+ )
390
+
391
+ output = json.loads(response["Body"].read().decode())
392
+
393
+ result = output["generated_text"]
394
+ return result
395
+
396
+ # please insert your long prompt/document content here
397
+ prompt = """<s>[INST] What are the main challenges to support long contexts for a Large Language Model? [/INST]"""
398
+
399
+ #print(prompt)
400
+ endpoint_name = "megaBeam-mistral-7b-300k-2024-05-13-14-23-41-219" # please use a valid endpoint name
401
+ result = call_endpoint(prompt, endpoint_name)
402
+ print(result)
403
+ ```
404
+
405
+
406
+ ## Limitations ##
407
+ Before using the MegaBeam-Mistral-7B-300k model, it is important to perform your own independent assessment, and take measures to ensure that your use would comply with your own specific quality control practices and standards, and that your use would comply with the local rules, laws, regulations, licenses and terms that apply to you, and your content.
408
+
409
+ ## The AWS Contributors ##
410
+ Chen Wu, Yin Song, Verdi March, Eden Duthie
config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "amazon/MegaBeam-Mistral-7B-300k",
3
+ "architectures": [
4
+ "MistralForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 14336,
13
+ "max_position_embeddings": 288800,
14
+ "model_type": "mistral",
15
+ "num_attention_heads": 32,
16
+ "num_hidden_layers": 32,
17
+ "num_key_value_heads": 8,
18
+ "quantization_config": {
19
+ "activation_scheme": "static",
20
+ "ignored_layers": [
21
+ "lm_head"
22
+ ],
23
+ "quant_method": "fp8"
24
+ },
25
+ "rms_norm_eps": 1e-05,
26
+ "rope_theta": 25000000.0,
27
+ "sliding_window": null,
28
+ "tie_word_embeddings": false,
29
+ "torch_dtype": "float16",
30
+ "transformers_version": "4.42.3",
31
+ "use_cache": false,
32
+ "vocab_size": 32000
33
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.42.3"
6
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b855f18e3f53965a71236bd703c5cd639cefd39fa1c9986cbb2d0d1fb29fbda
3
+ size 4943385232
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c9963f1644f9b253e2b50912947db42a0513cad6b0caf73b0d80e46b4449558
3
+ size 2560836920
model.safetensors.index.json ADDED
@@ -0,0 +1,746 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 7504144128
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "model-00002-of-00002.safetensors",
7
+ "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
8
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
9
+ "model.layers.0.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
10
+ "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
11
+ "model.layers.0.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
12
+ "model.layers.0.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
13
+ "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
14
+ "model.layers.0.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
15
+ "model.layers.0.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
16
+ "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
17
+ "model.layers.0.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
18
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
19
+ "model.layers.0.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
20
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
21
+ "model.layers.0.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
22
+ "model.layers.0.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
23
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
24
+ "model.layers.0.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
25
+ "model.layers.0.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
26
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
27
+ "model.layers.0.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
28
+ "model.layers.0.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
29
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
30
+ "model.layers.0.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
31
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
32
+ "model.layers.1.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
33
+ "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
34
+ "model.layers.1.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
35
+ "model.layers.1.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
36
+ "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
37
+ "model.layers.1.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
38
+ "model.layers.1.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
39
+ "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
40
+ "model.layers.1.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
41
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
42
+ "model.layers.1.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
43
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
44
+ "model.layers.1.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
45
+ "model.layers.1.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
46
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
47
+ "model.layers.1.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
48
+ "model.layers.1.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
49
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
50
+ "model.layers.1.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
51
+ "model.layers.1.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
52
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
53
+ "model.layers.1.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
54
+ "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
55
+ "model.layers.10.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
56
+ "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
57
+ "model.layers.10.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
58
+ "model.layers.10.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
59
+ "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
60
+ "model.layers.10.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
61
+ "model.layers.10.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
62
+ "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
63
+ "model.layers.10.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
64
+ "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
65
+ "model.layers.10.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
66
+ "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
67
+ "model.layers.10.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
68
+ "model.layers.10.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
69
+ "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
70
+ "model.layers.10.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
71
+ "model.layers.10.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
72
+ "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
73
+ "model.layers.10.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
74
+ "model.layers.10.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
75
+ "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
76
+ "model.layers.10.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
77
+ "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
78
+ "model.layers.11.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
79
+ "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
80
+ "model.layers.11.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
81
+ "model.layers.11.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
82
+ "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
83
+ "model.layers.11.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
84
+ "model.layers.11.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
85
+ "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
86
+ "model.layers.11.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
87
+ "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
88
+ "model.layers.11.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
89
+ "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
90
+ "model.layers.11.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
91
+ "model.layers.11.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
92
+ "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
93
+ "model.layers.11.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
94
+ "model.layers.11.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
95
+ "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
96
+ "model.layers.11.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
97
+ "model.layers.11.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
98
+ "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
99
+ "model.layers.11.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
100
+ "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
101
+ "model.layers.12.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
102
+ "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
103
+ "model.layers.12.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
104
+ "model.layers.12.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
105
+ "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
106
+ "model.layers.12.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
107
+ "model.layers.12.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
108
+ "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
109
+ "model.layers.12.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
110
+ "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
111
+ "model.layers.12.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
112
+ "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
113
+ "model.layers.12.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
114
+ "model.layers.12.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
115
+ "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
116
+ "model.layers.12.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
117
+ "model.layers.12.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
118
+ "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
119
+ "model.layers.12.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
120
+ "model.layers.12.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
121
+ "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
122
+ "model.layers.12.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
123
+ "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
124
+ "model.layers.13.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
125
+ "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
126
+ "model.layers.13.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
127
+ "model.layers.13.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
128
+ "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
129
+ "model.layers.13.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
130
+ "model.layers.13.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
131
+ "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
132
+ "model.layers.13.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
133
+ "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
134
+ "model.layers.13.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
135
+ "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
136
+ "model.layers.13.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
137
+ "model.layers.13.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
138
+ "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
139
+ "model.layers.13.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
140
+ "model.layers.13.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
141
+ "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
142
+ "model.layers.13.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
143
+ "model.layers.13.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
144
+ "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
145
+ "model.layers.13.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
146
+ "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
147
+ "model.layers.14.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
148
+ "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
149
+ "model.layers.14.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
150
+ "model.layers.14.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
151
+ "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
152
+ "model.layers.14.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
153
+ "model.layers.14.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
154
+ "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
155
+ "model.layers.14.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
156
+ "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
157
+ "model.layers.14.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
158
+ "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
159
+ "model.layers.14.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
160
+ "model.layers.14.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
161
+ "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
162
+ "model.layers.14.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
163
+ "model.layers.14.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
164
+ "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
165
+ "model.layers.14.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
166
+ "model.layers.14.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
167
+ "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
168
+ "model.layers.14.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
169
+ "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
170
+ "model.layers.15.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
171
+ "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
172
+ "model.layers.15.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
173
+ "model.layers.15.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
174
+ "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
175
+ "model.layers.15.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
176
+ "model.layers.15.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
177
+ "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
178
+ "model.layers.15.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
179
+ "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
180
+ "model.layers.15.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
181
+ "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
182
+ "model.layers.15.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
183
+ "model.layers.15.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
184
+ "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
185
+ "model.layers.15.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
186
+ "model.layers.15.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
187
+ "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
188
+ "model.layers.15.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
189
+ "model.layers.15.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
190
+ "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
191
+ "model.layers.15.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
192
+ "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
193
+ "model.layers.16.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
194
+ "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
195
+ "model.layers.16.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
196
+ "model.layers.16.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
197
+ "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
198
+ "model.layers.16.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
199
+ "model.layers.16.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
200
+ "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
201
+ "model.layers.16.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
202
+ "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
203
+ "model.layers.16.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
204
+ "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
205
+ "model.layers.16.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
206
+ "model.layers.16.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
207
+ "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
208
+ "model.layers.16.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
209
+ "model.layers.16.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
210
+ "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
211
+ "model.layers.16.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
212
+ "model.layers.16.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
213
+ "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
214
+ "model.layers.16.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
215
+ "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
216
+ "model.layers.17.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
217
+ "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
218
+ "model.layers.17.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
219
+ "model.layers.17.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
220
+ "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
221
+ "model.layers.17.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
222
+ "model.layers.17.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
223
+ "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
224
+ "model.layers.17.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
225
+ "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
226
+ "model.layers.17.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
227
+ "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
228
+ "model.layers.17.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
229
+ "model.layers.17.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
230
+ "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
231
+ "model.layers.17.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
232
+ "model.layers.17.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
233
+ "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
234
+ "model.layers.17.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
235
+ "model.layers.17.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
236
+ "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
237
+ "model.layers.17.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
238
+ "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
239
+ "model.layers.18.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
240
+ "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
241
+ "model.layers.18.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
242
+ "model.layers.18.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
243
+ "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
244
+ "model.layers.18.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
245
+ "model.layers.18.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
246
+ "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
247
+ "model.layers.18.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
248
+ "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
249
+ "model.layers.18.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
250
+ "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
251
+ "model.layers.18.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
252
+ "model.layers.18.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
253
+ "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
254
+ "model.layers.18.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
255
+ "model.layers.18.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
256
+ "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
257
+ "model.layers.18.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
258
+ "model.layers.18.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
259
+ "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
260
+ "model.layers.18.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
261
+ "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
262
+ "model.layers.19.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
263
+ "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
264
+ "model.layers.19.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
265
+ "model.layers.19.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
266
+ "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
267
+ "model.layers.19.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
268
+ "model.layers.19.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
269
+ "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
270
+ "model.layers.19.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
271
+ "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
272
+ "model.layers.19.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
273
+ "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
274
+ "model.layers.19.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
275
+ "model.layers.19.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
276
+ "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
277
+ "model.layers.19.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
278
+ "model.layers.19.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
279
+ "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
280
+ "model.layers.19.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
281
+ "model.layers.19.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
282
+ "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
283
+ "model.layers.19.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
284
+ "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
285
+ "model.layers.2.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
286
+ "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
287
+ "model.layers.2.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
288
+ "model.layers.2.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
289
+ "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
290
+ "model.layers.2.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
291
+ "model.layers.2.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
292
+ "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
293
+ "model.layers.2.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
294
+ "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
295
+ "model.layers.2.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
296
+ "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
297
+ "model.layers.2.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
298
+ "model.layers.2.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
299
+ "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
300
+ "model.layers.2.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
301
+ "model.layers.2.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
302
+ "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
303
+ "model.layers.2.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
304
+ "model.layers.2.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
305
+ "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
306
+ "model.layers.2.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
307
+ "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
308
+ "model.layers.20.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
309
+ "model.layers.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
310
+ "model.layers.20.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
311
+ "model.layers.20.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
312
+ "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
313
+ "model.layers.20.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
314
+ "model.layers.20.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
315
+ "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
316
+ "model.layers.20.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
317
+ "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
318
+ "model.layers.20.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
319
+ "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
320
+ "model.layers.20.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
321
+ "model.layers.20.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
322
+ "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
323
+ "model.layers.20.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
324
+ "model.layers.20.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
325
+ "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
326
+ "model.layers.20.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
327
+ "model.layers.20.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
328
+ "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
329
+ "model.layers.20.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
330
+ "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
331
+ "model.layers.21.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
332
+ "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
333
+ "model.layers.21.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
334
+ "model.layers.21.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
335
+ "model.layers.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
336
+ "model.layers.21.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
337
+ "model.layers.21.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
338
+ "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
339
+ "model.layers.21.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
340
+ "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
341
+ "model.layers.21.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
342
+ "model.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
343
+ "model.layers.21.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
344
+ "model.layers.21.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
345
+ "model.layers.21.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
346
+ "model.layers.21.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
347
+ "model.layers.21.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
348
+ "model.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
349
+ "model.layers.21.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
350
+ "model.layers.21.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
351
+ "model.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
352
+ "model.layers.21.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
353
+ "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
354
+ "model.layers.22.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
355
+ "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
356
+ "model.layers.22.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
357
+ "model.layers.22.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
358
+ "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
359
+ "model.layers.22.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
360
+ "model.layers.22.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
361
+ "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
362
+ "model.layers.22.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
363
+ "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
364
+ "model.layers.22.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
365
+ "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
366
+ "model.layers.22.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
367
+ "model.layers.22.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
368
+ "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
369
+ "model.layers.22.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
370
+ "model.layers.22.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
371
+ "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
372
+ "model.layers.22.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
373
+ "model.layers.22.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
374
+ "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
375
+ "model.layers.22.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
376
+ "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
377
+ "model.layers.23.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
378
+ "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
379
+ "model.layers.23.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
380
+ "model.layers.23.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
381
+ "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
382
+ "model.layers.23.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
383
+ "model.layers.23.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
384
+ "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
385
+ "model.layers.23.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
386
+ "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
387
+ "model.layers.23.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
388
+ "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
389
+ "model.layers.23.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
390
+ "model.layers.23.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
391
+ "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
392
+ "model.layers.23.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
393
+ "model.layers.23.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
394
+ "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
395
+ "model.layers.23.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
396
+ "model.layers.23.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
397
+ "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
398
+ "model.layers.23.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
399
+ "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
400
+ "model.layers.24.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
401
+ "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
402
+ "model.layers.24.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
403
+ "model.layers.24.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
404
+ "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
405
+ "model.layers.24.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
406
+ "model.layers.24.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
407
+ "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
408
+ "model.layers.24.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
409
+ "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
410
+ "model.layers.24.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
411
+ "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
412
+ "model.layers.24.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
413
+ "model.layers.24.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
414
+ "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
415
+ "model.layers.24.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
416
+ "model.layers.24.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
417
+ "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
418
+ "model.layers.24.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
419
+ "model.layers.24.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
420
+ "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
421
+ "model.layers.24.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
422
+ "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
423
+ "model.layers.25.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
424
+ "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
425
+ "model.layers.25.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
426
+ "model.layers.25.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
427
+ "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
428
+ "model.layers.25.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
429
+ "model.layers.25.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
430
+ "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
431
+ "model.layers.25.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
432
+ "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
433
+ "model.layers.25.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
434
+ "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
435
+ "model.layers.25.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
436
+ "model.layers.25.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
437
+ "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
438
+ "model.layers.25.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
439
+ "model.layers.25.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
440
+ "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
441
+ "model.layers.25.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
442
+ "model.layers.25.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
443
+ "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
444
+ "model.layers.25.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
445
+ "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
446
+ "model.layers.26.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
447
+ "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
448
+ "model.layers.26.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
449
+ "model.layers.26.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
450
+ "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
451
+ "model.layers.26.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
452
+ "model.layers.26.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
453
+ "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
454
+ "model.layers.26.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
455
+ "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
456
+ "model.layers.26.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
457
+ "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
458
+ "model.layers.26.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
459
+ "model.layers.26.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
460
+ "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
461
+ "model.layers.26.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
462
+ "model.layers.26.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
463
+ "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
464
+ "model.layers.26.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
465
+ "model.layers.26.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
466
+ "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
467
+ "model.layers.26.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
468
+ "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
469
+ "model.layers.27.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
470
+ "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
471
+ "model.layers.27.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
472
+ "model.layers.27.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
473
+ "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
474
+ "model.layers.27.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
475
+ "model.layers.27.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
476
+ "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
477
+ "model.layers.27.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
478
+ "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
479
+ "model.layers.27.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
480
+ "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
481
+ "model.layers.27.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
482
+ "model.layers.27.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
483
+ "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
484
+ "model.layers.27.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
485
+ "model.layers.27.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
486
+ "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
487
+ "model.layers.27.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
488
+ "model.layers.27.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
489
+ "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
490
+ "model.layers.27.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
491
+ "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
492
+ "model.layers.28.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
493
+ "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
494
+ "model.layers.28.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
495
+ "model.layers.28.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
496
+ "model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
497
+ "model.layers.28.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
498
+ "model.layers.28.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
499
+ "model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
500
+ "model.layers.28.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
501
+ "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
502
+ "model.layers.28.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
503
+ "model.layers.28.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
504
+ "model.layers.28.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
505
+ "model.layers.28.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
506
+ "model.layers.28.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
507
+ "model.layers.28.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
508
+ "model.layers.28.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
509
+ "model.layers.28.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
510
+ "model.layers.28.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
511
+ "model.layers.28.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
512
+ "model.layers.28.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
513
+ "model.layers.28.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
514
+ "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
515
+ "model.layers.29.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
516
+ "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
517
+ "model.layers.29.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
518
+ "model.layers.29.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
519
+ "model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
520
+ "model.layers.29.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
521
+ "model.layers.29.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
522
+ "model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
523
+ "model.layers.29.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
524
+ "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
525
+ "model.layers.29.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
526
+ "model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
527
+ "model.layers.29.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
528
+ "model.layers.29.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
529
+ "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
530
+ "model.layers.29.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
531
+ "model.layers.29.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
532
+ "model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
533
+ "model.layers.29.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
534
+ "model.layers.29.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
535
+ "model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
536
+ "model.layers.29.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
537
+ "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
538
+ "model.layers.3.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
539
+ "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
540
+ "model.layers.3.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
541
+ "model.layers.3.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
542
+ "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
543
+ "model.layers.3.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
544
+ "model.layers.3.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
545
+ "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
546
+ "model.layers.3.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
547
+ "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
548
+ "model.layers.3.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
549
+ "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
550
+ "model.layers.3.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
551
+ "model.layers.3.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
552
+ "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
553
+ "model.layers.3.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
554
+ "model.layers.3.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
555
+ "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
556
+ "model.layers.3.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
557
+ "model.layers.3.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
558
+ "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
559
+ "model.layers.3.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
560
+ "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
561
+ "model.layers.30.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
562
+ "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
563
+ "model.layers.30.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
564
+ "model.layers.30.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
565
+ "model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
566
+ "model.layers.30.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
567
+ "model.layers.30.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
568
+ "model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
569
+ "model.layers.30.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
570
+ "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
571
+ "model.layers.30.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
572
+ "model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
573
+ "model.layers.30.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
574
+ "model.layers.30.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
575
+ "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
576
+ "model.layers.30.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
577
+ "model.layers.30.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
578
+ "model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
579
+ "model.layers.30.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
580
+ "model.layers.30.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
581
+ "model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
582
+ "model.layers.30.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
583
+ "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
584
+ "model.layers.31.mlp.down_proj.input_scale": "model-00002-of-00002.safetensors",
585
+ "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
586
+ "model.layers.31.mlp.down_proj.weight_scale": "model-00002-of-00002.safetensors",
587
+ "model.layers.31.mlp.gate_proj.input_scale": "model-00002-of-00002.safetensors",
588
+ "model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
589
+ "model.layers.31.mlp.gate_proj.weight_scale": "model-00002-of-00002.safetensors",
590
+ "model.layers.31.mlp.up_proj.input_scale": "model-00002-of-00002.safetensors",
591
+ "model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
592
+ "model.layers.31.mlp.up_proj.weight_scale": "model-00002-of-00002.safetensors",
593
+ "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
594
+ "model.layers.31.self_attn.k_proj.input_scale": "model-00002-of-00002.safetensors",
595
+ "model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
596
+ "model.layers.31.self_attn.k_proj.weight_scale": "model-00002-of-00002.safetensors",
597
+ "model.layers.31.self_attn.o_proj.input_scale": "model-00002-of-00002.safetensors",
598
+ "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
599
+ "model.layers.31.self_attn.o_proj.weight_scale": "model-00002-of-00002.safetensors",
600
+ "model.layers.31.self_attn.q_proj.input_scale": "model-00002-of-00002.safetensors",
601
+ "model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
602
+ "model.layers.31.self_attn.q_proj.weight_scale": "model-00002-of-00002.safetensors",
603
+ "model.layers.31.self_attn.v_proj.input_scale": "model-00002-of-00002.safetensors",
604
+ "model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
605
+ "model.layers.31.self_attn.v_proj.weight_scale": "model-00002-of-00002.safetensors",
606
+ "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
607
+ "model.layers.4.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
608
+ "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
609
+ "model.layers.4.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
610
+ "model.layers.4.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
611
+ "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
612
+ "model.layers.4.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
613
+ "model.layers.4.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
614
+ "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
615
+ "model.layers.4.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
616
+ "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
617
+ "model.layers.4.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
618
+ "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
619
+ "model.layers.4.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
620
+ "model.layers.4.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
621
+ "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
622
+ "model.layers.4.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
623
+ "model.layers.4.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
624
+ "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
625
+ "model.layers.4.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
626
+ "model.layers.4.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
627
+ "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
628
+ "model.layers.4.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
629
+ "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
630
+ "model.layers.5.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
631
+ "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
632
+ "model.layers.5.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
633
+ "model.layers.5.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
634
+ "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
635
+ "model.layers.5.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
636
+ "model.layers.5.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
637
+ "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
638
+ "model.layers.5.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
639
+ "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
640
+ "model.layers.5.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
641
+ "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
642
+ "model.layers.5.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
643
+ "model.layers.5.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
644
+ "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
645
+ "model.layers.5.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
646
+ "model.layers.5.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
647
+ "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
648
+ "model.layers.5.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
649
+ "model.layers.5.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
650
+ "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
651
+ "model.layers.5.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
652
+ "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
653
+ "model.layers.6.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
654
+ "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
655
+ "model.layers.6.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
656
+ "model.layers.6.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
657
+ "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
658
+ "model.layers.6.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
659
+ "model.layers.6.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
660
+ "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
661
+ "model.layers.6.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
662
+ "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
663
+ "model.layers.6.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
664
+ "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
665
+ "model.layers.6.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
666
+ "model.layers.6.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
667
+ "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
668
+ "model.layers.6.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
669
+ "model.layers.6.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
670
+ "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
671
+ "model.layers.6.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
672
+ "model.layers.6.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
673
+ "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
674
+ "model.layers.6.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
675
+ "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
676
+ "model.layers.7.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
677
+ "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
678
+ "model.layers.7.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
679
+ "model.layers.7.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
680
+ "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
681
+ "model.layers.7.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
682
+ "model.layers.7.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
683
+ "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
684
+ "model.layers.7.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
685
+ "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
686
+ "model.layers.7.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
687
+ "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
688
+ "model.layers.7.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
689
+ "model.layers.7.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
690
+ "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
691
+ "model.layers.7.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
692
+ "model.layers.7.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
693
+ "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
694
+ "model.layers.7.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
695
+ "model.layers.7.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
696
+ "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
697
+ "model.layers.7.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
698
+ "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
699
+ "model.layers.8.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
700
+ "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
701
+ "model.layers.8.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
702
+ "model.layers.8.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
703
+ "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
704
+ "model.layers.8.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
705
+ "model.layers.8.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
706
+ "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
707
+ "model.layers.8.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
708
+ "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
709
+ "model.layers.8.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
710
+ "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
711
+ "model.layers.8.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
712
+ "model.layers.8.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
713
+ "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
714
+ "model.layers.8.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
715
+ "model.layers.8.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
716
+ "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
717
+ "model.layers.8.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
718
+ "model.layers.8.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
719
+ "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
720
+ "model.layers.8.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
721
+ "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
722
+ "model.layers.9.mlp.down_proj.input_scale": "model-00001-of-00002.safetensors",
723
+ "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
724
+ "model.layers.9.mlp.down_proj.weight_scale": "model-00001-of-00002.safetensors",
725
+ "model.layers.9.mlp.gate_proj.input_scale": "model-00001-of-00002.safetensors",
726
+ "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
727
+ "model.layers.9.mlp.gate_proj.weight_scale": "model-00001-of-00002.safetensors",
728
+ "model.layers.9.mlp.up_proj.input_scale": "model-00001-of-00002.safetensors",
729
+ "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
730
+ "model.layers.9.mlp.up_proj.weight_scale": "model-00001-of-00002.safetensors",
731
+ "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
732
+ "model.layers.9.self_attn.k_proj.input_scale": "model-00001-of-00002.safetensors",
733
+ "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
734
+ "model.layers.9.self_attn.k_proj.weight_scale": "model-00001-of-00002.safetensors",
735
+ "model.layers.9.self_attn.o_proj.input_scale": "model-00001-of-00002.safetensors",
736
+ "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
737
+ "model.layers.9.self_attn.o_proj.weight_scale": "model-00001-of-00002.safetensors",
738
+ "model.layers.9.self_attn.q_proj.input_scale": "model-00001-of-00002.safetensors",
739
+ "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
740
+ "model.layers.9.self_attn.q_proj.weight_scale": "model-00001-of-00002.safetensors",
741
+ "model.layers.9.self_attn.v_proj.input_scale": "model-00001-of-00002.safetensors",
742
+ "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
743
+ "model.layers.9.self_attn.v_proj.weight_scale": "model-00001-of-00002.safetensors",
744
+ "model.norm.weight": "model-00002-of-00002.safetensors"
745
+ }
746
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "unk_token": {
17
+ "content": "<unk>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ }
23
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
tokenizer_config.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "additional_special_tokens": [],
32
+ "bos_token": "<s>",
33
+ "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token}}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
34
+ "clean_up_tokenization_spaces": false,
35
+ "eos_token": "</s>",
36
+ "legacy": true,
37
+ "model_max_length": 1000000000000000019884624838656,
38
+ "pad_token": null,
39
+ "sp_model_kwargs": {},
40
+ "spaces_between_special_tokens": false,
41
+ "tokenizer_class": "LlamaTokenizer",
42
+ "unk_token": "<unk>",
43
+ "use_default_system_prompt": false
44
+ }