Skip to content

Commit

Permalink
[Doc] fix the autoAWQ example (vllm-project#7937)
Browse files Browse the repository at this point in the history
  • Loading branch information
stas00 authored and root committed Aug 28, 2024
1 parent 1a01945 commit 418fb05
Showing 1 changed file with 12 additions and 8 deletions.
20 changes: 12 additions & 8 deletions docs/source/quantization/auto_awq.rst
Original file line number Diff line number Diff line change
Expand Up @@ -19,27 +19,31 @@ You can quantize your own models by installing AutoAWQ or picking one of the `40
$ pip install autoawq
After installing AutoAWQ, you are ready to quantize a model. Here is an example of how to quantize Vicuna 7B v1.5:
After installing AutoAWQ, you are ready to quantize a model. Here is an example of how to quantize `mistralai/Mistral-7B-Instruct-v0.2`:

.. code-block:: python
from awq import AutoAWQForCausalLM
from transformers import AutoTokenizer
model_path = 'lmsys/vicuna-7b-v1.5'
quant_path = 'vicuna-7b-v1.5-awq'
model_path = 'mistralai/Mistral-7B-Instruct-v0.2'
quant_path = 'mistral-instruct-v0.2-awq'
quant_config = { "zero_point": True, "q_group_size": 128, "w_bit": 4, "version": "GEMM" }
# Load model
model = AutoAWQForCausalLM.from_pretrained(model_path, **{"low_cpu_mem_usage": True})
model = AutoAWQForCausalLM.from_pretrained(
model_path, **{"low_cpu_mem_usage": True, "use_cache": False}
)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
# Quantize
model.quantize(tokenizer, quant_config=quant_config)
# Save quantized model
model.save_quantized(quant_path)
tokenizer.save_pretrained(quant_path)
print(f'Model is quantized and saved at "{quant_path}"')
To run an AWQ model with vLLM, you can use `TheBloke/Llama-2-7b-Chat-AWQ <https://huggingface.co/TheBloke/Llama-2-7b-Chat-AWQ>`_ with the following command:

Expand Down

0 comments on commit 418fb05

Please sign in to comment.