|
--- |
|
tags: |
|
- fp8 |
|
- vllm |
|
license: apache-2.0 |
|
--- |
|
|
|
# Qwen2-1.5B-Instruct-FP8 |
|
|
|
## Model Overview |
|
- **Model Architecture:** Qwen2 |
|
- **Input:** Text |
|
- **Output:** Text |
|
- **Model Optimizations:** |
|
- **Weight quantization:** FP8 |
|
- **Activation quantization:** FP8 |
|
- **Intended Use Cases:** Intended for commercial and research use in English. Similarly to [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct), this models is intended for assistant-like chat. |
|
- **Out-of-scope:** Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English. |
|
- **Release Date:** 6/14/2024 |
|
- **Version:** 1.0 |
|
- **License(s):** [apache-2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
|
- **Model Developers:** Neural Magic |
|
|
|
Quantized version of [Qwen2-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2-1.5B-Instruct). |
|
It achieves an average score of 54.59 on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves 55.18. |
|
|
|
### Model Optimizations |
|
|
|
This model was obtained by quantizing the weights and activations of [Qwen2-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2-1.5B-Instruct) to FP8 data type, ready for inference with vLLM >= 0.5.0. |
|
This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%. |
|
|
|
Only the weights and activations of the linear operators within transformers blocks are quantized. Symmetric per-tensor quantization is applied, in which a single linear scaling maps the FP8 representations of the quantized weights and activations. |
|
[AutoFP8](https://github.com/neuralmagic/AutoFP8) is used for quantization with 512 sequences of UltraChat. |
|
|
|
## Deployment |
|
|
|
### Use with vLLM |
|
|
|
This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) backend, as shown in the example below. |
|
|
|
```python |
|
from vllm import LLM, SamplingParams |
|
from transformers import AutoTokenizer |
|
|
|
model_id = "neuralmagic/Qwen2-1.5B-Instruct-FP8" |
|
|
|
sampling_params = SamplingParams(temperature=0.6, top_p=0.9, max_tokens=256) |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id) |
|
|
|
messages = [ |
|
{"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, |
|
{"role": "user", "content": "Who are you?"}, |
|
] |
|
|
|
prompts = tokenizer.apply_chat_template(messages, tokenize=False) |
|
|
|
llm = LLM(model=model_id) |
|
|
|
outputs = llm.generate(prompts, sampling_params) |
|
|
|
generated_text = outputs[0].outputs[0].text |
|
print(generated_text) |
|
``` |
|
|
|
vLLM aslo supports OpenAI-compatible serving. See the [documentation](https://docs.vllm.ai/en/latest/) for more details. |
|
|
|
## Creation |
|
|
|
This model was created by applying [AutoFP8 with calibration samples from ultrachat](https://github.com/neuralmagic/AutoFP8/blob/147fa4d9e1a90ef8a93f96fc7d9c33056ddc017a/example_dataset.py), as presented in the code snipet below. |
|
Although AutoFP8 was used for this particular model, Neural Magic is transitioning to using [llm-compressor](https://github.com/vllm-project/llm-compressor) which supports several quantization schemes and models not supported by AutoFP8. |
|
|
|
```python |
|
from datasets import load_dataset |
|
from transformers import AutoTokenizer |
|
|
|
from auto_fp8 import AutoFP8ForCausalLM, BaseQuantizeConfig |
|
|
|
pretrained_model_dir = "Qwen/Qwen2-1.5B-Instruct" |
|
quantized_model_dir = "Qwen2-1.5B-Instruct-FP8" |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_dir, use_fast=True, model_max_length=4096) |
|
tokenizer.pad_token = tokenizer.eos_token |
|
|
|
ds = load_dataset("mgoin/ultrachat_2k", split="train_sft").select(range(512)) |
|
examples = [tokenizer.apply_chat_template(batch["messages"], tokenize=False) for batch in ds] |
|
examples = tokenizer(examples, padding=True, truncation=True, return_tensors="pt").to("cuda") |
|
|
|
quantize_config = BaseQuantizeConfig( |
|
quant_method="fp8", |
|
activation_scheme="static" |
|
ignore_patterns=["re:.*lm_head"], |
|
) |
|
|
|
model = AutoFP8ForCausalLM.from_pretrained( |
|
pretrained_model_dir, quantize_config=quantize_config |
|
) |
|
|
|
model.quantize(examples) |
|
model.save_quantized(quantized_model_dir) |
|
``` |
|
|
|
## Evaluation |
|
|
|
The model was evaluated on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) leaderboard tasks (version 1) with the [lm-evaluation-harness](https://github.com/EleutherAI/lm-evaluation-harness/tree/383bbd54bc621086e05aa1b030d8d4d5635b25e6) (commit 383bbd54bc621086e05aa1b030d8d4d5635b25e6) and the [vLLM](https://docs.vllm.ai/en/stable/) engine, using the following command: |
|
``` |
|
lm_eval \ |
|
--model vllm \ |
|
--model_args pretrained="neuralmagic/Qwen2-1.5B-Instruct-FP8",dtype=auto,gpu_memory_utilization=0.4,add_bos_token=True,max_model_len=4096 \ |
|
--tasks openllm \ |
|
--batch_size auto |
|
``` |
|
|
|
### Accuracy |
|
|
|
#### Open LLM Leaderboard evaluation scores |
|
<table> |
|
<tr> |
|
<td><strong>Benchmark</strong> |
|
</td> |
|
<td><strong>Qwen2-1.5B-Instruct</strong> |
|
</td> |
|
<td><strong>Qwen2-1.5B-Instruct-FP8(this model)</strong> |
|
</td> |
|
<td><strong>Recovery</strong> |
|
</td> |
|
</tr> |
|
<tr> |
|
<td>MMLU (5-shot) |
|
</td> |
|
<td>55.87 |
|
</td> |
|
<td>55.60 |
|
</td> |
|
<td>99.51% |
|
</td> |
|
</tr> |
|
<tr> |
|
<td>ARC Challenge (25-shot) |
|
</td> |
|
<td>43.09 |
|
</td> |
|
<td>41.81 |
|
</td> |
|
<td>97.02% |
|
</td> |
|
</tr> |
|
<tr> |
|
<td>GSM-8K (5-shot, strict-match) |
|
</td> |
|
<td>57.70 |
|
</td> |
|
<td>56.48 |
|
</td> |
|
<td>97.88% |
|
</td> |
|
</tr> |
|
<tr> |
|
<td>Hellaswag (10-shot) |
|
</td> |
|
<td>67.48 |
|
</td> |
|
<td>67.18 |
|
</td> |
|
<td>99.55% |
|
</td> |
|
</tr> |
|
<tr> |
|
<td>Winogrande (5-shot) |
|
</td> |
|
<td>63.61 |
|
</td> |
|
<td>63.38 |
|
</td> |
|
<td>99.63% |
|
</td> |
|
</tr> |
|
<tr> |
|
<td>TruthfulQA (0-shot) |
|
</td> |
|
<td>43.34 |
|
</td> |
|
<td>43.09 |
|
</td> |
|
<td>99.42% |
|
</td> |
|
</tr> |
|
<tr> |
|
<td><strong>Average</strong> |
|
</td> |
|
<td><strong>55.18</strong> |
|
</td> |
|
<td><strong>54.59</strong> |
|
</td> |
|
<td><strong>98.93%</strong> |
|
</td> |
|
</tr> |
|
</table> |