metadata
language:
- en
tags:
- pytorch
- causal-lm
- olmo
- autoround
- intel-autoround
- gptq
- woq
- intel
- pytorch
- allenai
license: apache-2.0
model_name: OLMo 2 7B
base_model:
- allenai/OLMo-2-1124-7B
inference: false
model_creator: allenai
datasets:
- allenai/RLVR-GSM
pipeline_tag: text-generation
prompt_template: '{prompt} '
quantized_by: fbaldassarri
Model Information
Quantized version of allenai/OLMo-2-1124-7B using torch.float32 for quantization tuning.
- 4 bits (INT4)
- group size = 128
- Symmetrical Quantization
- Method WoQ (AutoRound format)
Fast and low memory, 2-3X speedup (slight accuracy drop at W4G128)
Quantization framework: Intel AutoRound v0.4.2
Note: this INT4 version of OLMo-2-1124-7B has been quantized to run inference through CPU.
Replication Recipe
Step 1 Install Requirements
I suggest to install requirements into a dedicated python-virtualenv or a conda enviroment.
wget https://github.com/intel/auto-round/archive/refs/tags/v0.4.2.tar.gz
tar -xvzf v0.4.2.tar.gz
cd auto-round-0.4.2
pip install -r requirements-cpu.txt --upgrade
Step 2 Build Intel AutoRound wheel from sources
pip install -vvv --no-build-isolation -e .[cpu]
Step 3 Script for Quantization
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "allenai/OLMo-2-1124-7B"
model = AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)
from auto_round import AutoRound
bits, group_size, sym, device, amp = 4, 128, True, 'cpu', False
autoround = AutoRound(model, tokenizer, nsamples=128, iters=200, seqlen=512, batch_size=4, bits=bits, group_size=group_size, sym=sym, device=device, amp=amp)
autoround.quantize()
output_dir = "./AutoRound/allenai_OLMo-2-1124-7B-autoround-int4-gs128-sym"
autoround.save_quantized(output_dir, format='auto_round', inplace=True)
License
Disclaimer
This quantized model comes with no warranty. It has been developed only for research purposes.