Instructions to use 4bit/llama-13b-3bit-gr128 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 4bit/llama-13b-3bit-gr128 with Transformers:

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

pipe = pipeline("text-generation", model="4bit/llama-13b-3bit-gr128")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("4bit/llama-13b-3bit-gr128")
model = AutoModelForCausalLM.from_pretrained("4bit/llama-13b-3bit-gr128")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use 4bit/llama-13b-3bit-gr128 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "4bit/llama-13b-3bit-gr128"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "4bit/llama-13b-3bit-gr128",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/4bit/llama-13b-3bit-gr128

SGLang

How to use 4bit/llama-13b-3bit-gr128 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 "4bit/llama-13b-3bit-gr128" \
    --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": "4bit/llama-13b-3bit-gr128",
		"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 "4bit/llama-13b-3bit-gr128" \
        --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": "4bit/llama-13b-3bit-gr128",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use 4bit/llama-13b-3bit-gr128 with Docker Model Runner:
```
docker model run hf.co/4bit/llama-13b-3bit-gr128
```

camenduru commited on Apr 9, 2023

Commit

bb9debd

1 Parent(s): 3b208f0

thanks to wcde ❤

Browse files

Files changed (7) hide show

README.md +1 -0
config.json +17 -0
generation_config.json +5 -0
llama-13b-3bit-gr128.pt +3 -0
special_tokens_map.json +1 -0
tokenizer.model +3 -0
tokenizer_config.json +5 -0

README.md ADDED Viewed

	@@ -0,0 +1 @@


1	+ Generated with: --wbits 3 --groupsize 128 --true-sequential --new-eval --faster-kernel

config.json ADDED Viewed

	@@ -0,0 +1,17 @@

+{"architectures": ["LLaMAForCausalLM"],
+"bos_token_id": 0,
+"eos_token_id": 1,
+"hidden_act": "silu",
+"hidden_size": 5120,
+"intermediate_size": 13824,
+"initializer_range": 0.02,
+"max_sequence_length": 2048,
+"model_type": "llama",
+"num_attention_heads": 40,
+"num_hidden_layers": 40,
+"pad_token_id": -1,
+"rms_norm_eps": 1e-06,
+"torch_dtype": "float16",
+"transformers_version": "4.27.0.dev0",
+"use_cache": true,
+"vocab_size": 32000}

generation_config.json ADDED Viewed

	@@ -0,0 +1,5 @@

+{"_from_model_config": true,
+"bos_token_id": 0,
+"eos_token_id": 1,
+"pad_token_id": 0,
+"transformers_version": "4.27.0.dev0"}

llama-13b-3bit-gr128.pt ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:e8d2cb21bd3a9c14e2cad1f6f8ab2179c04a29e8e505acae957d129980dea933
+size 5856740207

special_tokens_map.json ADDED Viewed

	@@ -0,0 +1 @@


1	+ {}

tokenizer.model ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json ADDED Viewed

	@@ -0,0 +1,5 @@

+{"bos_token": "",
+"eos_token": "",
+"model_max_length": 1000000000000000019884624838656,
+"tokenizer_class": "LlamaTokenizer",
+"unk_token": ""}