Instructions to use INC4AI/gpt-j-6b-sparse with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use INC4AI/gpt-j-6b-sparse with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="INC4AI/gpt-j-6b-sparse")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("INC4AI/gpt-j-6b-sparse") model = AutoModelForCausalLM.from_pretrained("INC4AI/gpt-j-6b-sparse", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use INC4AI/gpt-j-6b-sparse with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "INC4AI/gpt-j-6b-sparse" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "INC4AI/gpt-j-6b-sparse", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/INC4AI/gpt-j-6b-sparse
- SGLang
How to use INC4AI/gpt-j-6b-sparse 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 "INC4AI/gpt-j-6b-sparse" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "INC4AI/gpt-j-6b-sparse", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "INC4AI/gpt-j-6b-sparse" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "INC4AI/gpt-j-6b-sparse", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use INC4AI/gpt-j-6b-sparse with Docker Model Runner:
docker model run hf.co/INC4AI/gpt-j-6b-sparse
| language: | |
| - en | |
| tags: | |
| - pytorch | |
| - causal-lm | |
| license: apache-2.0 | |
| # Sparse GPT-J 6B | |
| ## Model Description | |
| The sparse version of GPT-J 6B is a pruned variant derived from the original [GPT-J 6B](https://huggingface.co/EleutherAI/gpt-j-6b) model and the vast majority of linear layers maintain a 40% unstructured sparsity (except for the 'lm_head'). | |
| <figure> | |
| | Hyperparameter | Value | | |
| |----------------------|------------| | |
| | \\(n_{parameters}\\) | 6053381344 | | |
| | \\(n_{layers}\\) | 28* | | |
| | \\(d_{model}\\) | 4096 | | |
| | \\(d_{ff}\\) | 16384 | | |
| | \\(n_{heads}\\) | 16 | | |
| | \\(d_{head}\\) | 256 | | |
| | \\(n_{ctx}\\) | 2048 | | |
| | \\(n_{vocab}\\) | 50257/50400† (same tokenizer as GPT-2/3) | | |
| | Positional Encoding | Rotary Position Embedding RoPE | | |
| | RoPE Dimensions | [64](https://github.com/kingoflolz/mesh-transformer-jax/blob/f2aa66e0925de6593dcbb70e72399b97b4130482/mesh_transformer/layers.py#L223) | | |
| <figcaption><p><strong>*</strong> Each layer consists of one feedforward block and one self attention block.</p> | |
| <p><strong>†</strong> Although the embedding matrix has a size of 50400, only 50257 entries are used by the GPT-2 tokenizer.</p></figcaption></figure> | |
| The model consists of 28 layers with a model dimension of 4096, and a feedforward dimension of 16384. The model | |
| dimension is split into 16 heads, each with a dimension of 256. Rotary Position Embedding (RoPE) is applied to 64 | |
| dimensions of each head. The model is trained with a tokenization vocabulary of 50257, using the same set of BPEs as | |
| GPT-2/GPT-3. | |
| ## Evaluation results | |
| Evaluating the accuracy of the sparse model of gpt-j-6b using the lambada_openai dataset in lm_eval, providing the accuracy fluctuation under two precisions: FP32 and BF16. | |
| <figure> | |
| | Sparsity | Dataset | Precision | Dense Acc ↑ | Sparse Acc ↑ | Acc fluctuations | | |
| |------ |---------------- |------- |------- |-------- |------------------ | | |
| | 40% |Lambada_openai | FP32 | 0.6831 | 0.6922 | +1.33% | | |
| | 40% |Lambada_openai | BF16 | 0.6771 | 0.6874 | +0.63% | |