Upload folder using huggingface_hub

README.md

@@ -10,87 +10,91 @@ This is the senior data synthesis model of SPEED.
 
 ## Usage
 
-Below is an example to synthesize classification data using this senior generator.
+Below is an example to synthesize classification data using this senior generator. 
+
+The prompts and misc scripts can be found in our [github page](https://github.com/haon-chen/SPEED)
 
 ### Transformers
 
 ```python
 import torch
-import torch.nn.functional as F
+import os
+import random
+import numpy as np
+import json
+import re
 
 from torch import Tensor
-from transformers import AutoTokenizer, AutoModel
-
-
-def last_token_pool(last_hidden_states: Tensor,
-                 attention_mask: Tensor) -> Tensor:
-    left_padding = (attention_mask[:, -1].sum() == attention_mask.shape[0])
-    if left_padding:
-        return last_hidden_states[:, -1]
-    else:
-        sequence_lengths = attention_mask.sum(dim=1) - 1
-        batch_size = last_hidden_states.shape[0]
-        return last_hidden_states[torch.arange(batch_size, device=last_hidden_states.device), sequence_lengths]
+from transformers import AutoTokenizer, AutoModelForCausalLM
 
+from prompts_synthesis import get_create_classify_data_prompt
+from utils import fix_common_json_errors_and_loads
 
-def get_detailed_instruct(task_description: str, query: str) -> str:
-    return f'Instruct: {task_description}\nQuery: {query}'
 
+LLAMA3_PROMPT = """
+{prompt} [/INST]
+""".strip("\n")
 
 # Each query must come with a one-sentence instruction that describes the task
-task = 'Given a web search query, retrieve relevant passages that answer the query'
-queries = [
-    get_detailed_instruct(task, 'how much protein should a female eat'),
-    get_detailed_instruct(task, 'summit define')
+tasks = [
+    'Identify the intended age group for educational technology products.',
+    'Classify businesses based on their operational hours.'
 ]
-# No need to add instruction for retrieval documents
-documents = [
-    "As a general guideline, the CDC's average requirement of protein for women ages 19 to 70 is 46 grams per day. But, as you can see from this chart, you'll need to increase that if you're expecting or training for a marathon. Check out the chart below to see how much protein you should be eating each day.",
-    "Definition of summit for English Language Learners. : 1  the highest point of a mountain : the top of a mountain. : 2  the highest level. : 3  a meeting or series of meetings between the leaders of two or more governments."
-]
-input_texts = queries + documents
-
-tokenizer = AutoTokenizer.from_pretrained('Haon-Chen/speed-embedding-7b-instruct')
-model = AutoModel.from_pretrained('Haon-Chen/speed-embedding-7b-instruct')
-
-max_length = 4096
-# Tokenize the input texts
-batch_dict = tokenizer(input_texts, max_length=max_length, padding=True, truncation=True, return_tensors='pt')
-
-outputs = model(**batch_dict)
-embeddings = last_token_pool(outputs.last_hidden_state, batch_dict['attention_mask'])
-
-# normalize embeddings
-embeddings = F.normalize(embeddings, p=2, dim=1)
-scores = (embeddings[:2] @ embeddings[2:].T) * 100
-print(scores.tolist())
+language = 'English'
+
+prompts = [LLAMA3_PROMPT.format(prompt=get_create_classify_data_prompt(task=task, language=language)[1]['content']) for task in tasks]
+
+tokenizer = AutoTokenizer.from_pretrained('Haon-Chen/speed-synthesis-7b-senior')
+model = AutoModelForCausalLM.from_pretrained('Haon-Chen/speed-synthesis-7b-senior')
+model.to("cuda:0")
+model.eval()
+tokenizer.pad_token = tokenizer.pad_token or tokenizer.eos_token
+tokenizer.padding_side = "left"
+tokenizer.truncation_side = "left"
+
+with torch.inference_mode():
+    # Tokenize the input texts
+    encodes = tokenizer(prompts, padding="longest", add_special_tokens=True, return_tensors="pt")
+    input_ids = encodes.input_ids.to(model.device)
+    attention_mask = encodes.attention_mask.to(model.device)
+
+    # Set the generation parameters
+    GEN_CONFIG = {"do_sample":True, "temperature": 1.0, "top_p": 1.0, "max_new_tokens": 800}
+    output = model.generate(
+        input_ids=input_ids,
+        attention_mask=attention_mask,
+        pad_token_id = tokenizer.eos_token_id,
+        **GEN_CONFIG
+    )
+output_texts = tokenizer.batch_decode(output, skip_special_tokens=True, clean_up_tokenization_spaces=False)
+batch_results = []
+for i in range(len(output_texts)):
+    batch_results.append(output_texts[i][len(prompts[i]):].strip(' '))
+
+# Format outputs
+bad_cnt=0
+outputs = []
+for i, result in enumerate(batch_results):
+    try:
+        output = fix_common_json_errors_and_loads(result)
+        user_query = output.get("input_text", "")
+        positive_document = output.get("label", "")
+        hard_negative_document = output.get("misleading_label", "")
+    except:
+        bad_cnt+=1
+        continue
+    out_data = {
+        "query": user_query,
+        "positives": [positive_document],
+        "negatives": [hard_negative_document],
+        "language": "English",
+        "task_definition": tasks[i],
+    }
+    outputs.append(out_data)
+print(bad_cnt)
+print(outputs)
 ```
 
-## MTEB Benchmark Evaluation
-
-Check out [unilm/e5](https://github.com/microsoft/unilm/tree/master/e5) to reproduce evaluation results 
-on the [BEIR](https://arxiv.org/abs/2104.08663) and [MTEB benchmark](https://arxiv.org/abs/2210.07316).
-
-## FAQ
-
-**1. Do I need to add instructions to the query?**
-
-Yes, this is how the model is trained, otherwise you will see a performance degradation.
-The task definition should be a one-sentence instruction that describes the task.
-This is a way to customize text embeddings for different scenarios through natural language instructions.
-
-Please check out [unilm/e5/utils.py](https://github.com/microsoft/unilm/blob/9c0f1ff7ca53431fe47d2637dfe253643d94185b/e5/utils.py#L106) for instructions we used for evaluation.
-
-On the other hand, there is no need to add instructions to the document side.
-
-**2. Why are my reproduced results slightly different from reported in the model card?**
-
-Different versions of `transformers` and `pytorch` could cause negligible but non-zero performance differences.
-
-**3. Where are the LoRA-only weights?**
-
-You can find the LoRA-only weights at [https://huggingface.co/Haon-Chen/speed-embedding-7b-instruct/tree/main/lora](https://huggingface.co/Haon-Chen/speed-embedding-7b-instruct/tree/main/lora).
-
 ## Citation
 
 If you find our paper or models helpful, please consider cite as follows:
@@ -105,5 +109,3 @@ If you find our paper or models helpful, please consider cite as follows:
 ```
 
 ## Limitations
-
-Using this model for inputs longer than 4096 tokens is not recommended.