Instructions to use GritLM/GritLM-8x7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use GritLM/GritLM-8x7B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="GritLM/GritLM-8x7B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("GritLM/GritLM-8x7B", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("GritLM/GritLM-8x7B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use GritLM/GritLM-8x7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "GritLM/GritLM-8x7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "GritLM/GritLM-8x7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/GritLM/GritLM-8x7B

SGLang

How to use GritLM/GritLM-8x7B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "GritLM/GritLM-8x7B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "GritLM/GritLM-8x7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "GritLM/GritLM-8x7B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "GritLM/GritLM-8x7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use GritLM/GritLM-8x7B with Docker Model Runner:
```
docker model run hf.co/GritLM/GritLM-8x7B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Model Summary

GritLM is a generative representational instruction tuned language model. It unifies text representation (embedding) and text generation into a single model achieving state-of-the-art performance on both types of tasks.

Repository: ContextualAI/gritlm
Paper: https://arxiv.org/abs/2402.09906
Logs: https://wandb.ai/muennighoff/gritlm/runs/id130s1m/overview
Script: https://github.com/ContextualAI/gritlm/blob/main/scripts/training/train_gritlm_8x7b.sh

Model	Description
GritLM 7B	Mistral 7B finetuned using GRIT
GritLM 8x7B	Mixtral 8x7B finetuned using GRIT

Use

The model usage is documented here.

Citation

@misc{muennighoff2024generative,
      title={Generative Representational Instruction Tuning}, 
      author={Niklas Muennighoff and Hongjin Su and Liang Wang and Nan Yang and Furu Wei and Tao Yu and Amanpreet Singh and Douwe Kiela},
      year={2024},
      eprint={2402.09906},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

Downloads last month: 8,353

Safetensors

Model size

47B params

Tensor type

BF16

Model tree for GritLM/GritLM-8x7B

Base model

mistralai/Mixtral-8x7B-v0.1

Finetuned

mistralai/Mixtral-8x7B-Instruct-v0.1

Finetuned

(65)

this model

Quantizations

4 models

Dataset used to train GritLM/GritLM-8x7B

Spaces using GritLM/GritLM-8x7B 15

Collection including GritLM/GritLM-8x7B

GritLM

Collection

Generative Representational Instruction Tuning (GRIT) • 63 items • Updated Mar 2 • 9

Paper for GritLM/GritLM-8x7B

Generative Representational Instruction Tuning

Paper • 2402.09906 • Published Feb 15, 2024 • 54

Evaluation results

accuracy on MTEB AmazonCounterfactualClassification (en)
test set self-reported

80.478
ap on MTEB AmazonCounterfactualClassification (en)
test set self-reported

44.388
f1 on MTEB AmazonCounterfactualClassification (en)
test set self-reported

74.336
accuracy on MTEB AmazonPolarityClassification
test set self-reported

96.322
ap on MTEB AmazonPolarityClassification
test set self-reported

94.803
f1 on MTEB AmazonPolarityClassification
test set self-reported

96.321
accuracy on MTEB AmazonReviewsClassification (en)
test set self-reported

57.184
f1 on MTEB AmazonReviewsClassification (en)
test set self-reported

55.945
map_at_1 on MTEB ArguAna
test set self-reported

34.353
map_at_10 on MTEB ArguAna
test set self-reported

50.773
map_at_100 on MTEB ArguAna
test set self-reported

51.515
map_at_1000 on MTEB ArguAna
test set self-reported

51.517
map_at_3 on MTEB ArguAna
test set self-reported

46.290
map_at_5 on MTEB ArguAna
test set self-reported

48.914
mrr_at_1 on MTEB ArguAna
test set self-reported

35.135
mrr_at_10 on MTEB ArguAna
test set self-reported

51.036