Model Details
This model is an int4 model with group_size128 and sym quantization of microsoft/phi-2 generated by intel/auto-round. We found there is a large accuracy drop of asym kernel for this model.
Use the model
INT4 Inference with ITREX on CPU
Install the latest intel-extension-for-transformers
from intel_extension_for_transformers.transformers import AutoModelForCausalLM
from transformers import AutoTokenizer
quantized_model_dir = "Intel/phi-2-int4-inc"
model = AutoModelForCausalLM.from_pretrained(quantized_model_dir,
device_map="auto",
trust_remote_code=False,
use_neural_speed=False,
)
tokenizer = AutoTokenizer.from_pretrained(quantized_model_dir, use_fast=True)
print(tokenizer.decode(model.generate(**tokenizer("There is a girl who likes adventure,", return_tensors="pt").to(model.device),max_new_tokens=50)[0]))
"""
There is a girl who likes adventure,
She loves to explore and to venture.
She travels to faraway lands,
And meets people from different lands.
She learns new languages and cultures,
And makes friends with all kinds of people.
She is curious and brave and
"""
INT4 Inference with AutoGPTQ
pip install auto-gptq
from transformers import AutoModelForCausalLM, AutoTokenizer
quantized_model_dir = "Intel/phi-2-int4-inc"
tokenizer = AutoTokenizer.from_pretrained(quantized_model_dir)
model = AutoModelForCausalLM.from_pretrained(quantized_model_dir, device_map="auto", trust_remote_code=True)
text = "There is a girl who likes adventure,"
inputs = tokenizer(text, return_tensors="pt", return_attention_mask=False).to(model.device)
outputs = model.generate(**inputs, max_new_tokens=50)
text = tokenizer.batch_decode(outputs)[0]
print(text)
"""
There is a girl who likes adventure,
She loves to explore and to venture.
She travels to faraway lands,
And meets people from different lands.
She learns new languages and cultures,
And makes friends with all kinds of people.
She is curious and brave and
"""
Evaluate the model
pip install lm-eval==0.4.2
lm_eval --model hf --model_args pretrained="Intel/phi-2-int4-inc" --device cuda:0 --tasks lambada_openai,hellaswag,piqa,winogrande,truthfulqa_mc1,openbookqa,boolq,arc_easy,arc_challenge,mmlu --batch_size 16
Metric | FP16 | INT4 |
---|---|---|
Avg. | 0.6131 | 0.6062 |
mmlu | 0.5334 | 0.5241 |
lambada_openai | 0.6243 | 0.6039 |
hellaswag | 0.5581 | 0.5487 |
winogrande | 0.7522 | 0.7585 |
piqa | 0.7867 | 0.7840 |
truthfulqa_mc1 | 0.3097 | 0.2974 |
openbookqa | 0.4040 | 0.3960 |
boolq | 0.8346 | 0.8346 |
arc_easy | 0.8001 | 0.8013 |
arc_challenge | 0.5282 | 0.5137 |
Reproduce the model
Here is the sample command to reproduce the model
git clone https://github.com/intel/auto-round
cd auto-round/examples/language-modeling
pip install -r requirements.txt
python3 main.py \
--model_name microsoft/phi-2 \
--device 0 \
--group_size 128 \
--bits 4 \
--iters 1000 \
--sym \
--nsamples 512 \
--deployment_device 'gpu' \
--output_dir "./tmp_autoround" \
Ethical Considerations and Limitations
The model can produce factually incorrect output, and should not be relied on to produce factually accurate information. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs.
Therefore, before deploying any applications of the model, developers should perform safety testing.
Caveats and Recommendations
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.
Here are a couple of useful links to learn more about Intel's AI software:
Disclaimer
The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please consult an attorney before using this model for commercial purposes.
- Downloads last month
- 11