DeepSeek-Coder-Lite

It achieves an average score of 79.60 on the HumanEval+ benchmark, whereas the unquantized model achieves 79.33.

Model Optimizations

This model was obtained by quantizing the weights and activations of DeepSeek-Coder-V2-Lite-Instruct to FP8 data type, ready for inference with vLLM >= 0.5.2. 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 is used for quantization with 512 sequences of UltraChat.

Deployment

Use with vLLM

This model can be deployed efficiently using the vLLM backend, as shown in the example below.

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "neuralmagic/DeepSeek-Coder-V2-Lite-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, add_generation_prompt=True)

llm = LLM(model=model_id, trust_remote_code=True, max_model_len=4096)

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 for more details.

Creation

This model was created by applying AutoFP8 with calibration samples from ultrachat with expert gates kept at original precision, as presented in the code snipet below. Although AutoFP8 was used for this particular model, Neural Magic is transitioning to using llm-compressor which supports several quantization schemes and models not supported by AutoFP8.

from datasets import load_dataset
from transformers import AutoTokenizer

from auto_fp8 import AutoFP8ForCausalLM, BaseQuantizeConfig

pretrained_model_dir = "deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct"
quantized_model_dir = "DeepSeek-Coder-V2-Lite-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 HumanEval+ benchmark with the Neural Magic fork of the EvalPlus implementation of HumanEval+ and the vLLM engine, using the following command:

python codegen/generate.py --model neuralmagic/DeepSeek-Coder-V2-Lite-Instruct-FP8 --temperature 0.2 --n_samples 50 --resume --root ~ --dataset humaneval
python evalplus/sanitize.py ~/humaneval/neuralmagic--DeepSeek-Coder-V2-Lite-Instruct-FP8_vllm_temp_0.2
evalplus.evaluate --dataset humaneval --samples ~/humaneval/neuralmagic--DeepSeek-Coder-V2-Lite-Instruct-FP8_vllm_temp_0.2-sanitized

Accuracy

HumanEval+ evaluation scores