neuralmagic / Phi-3.5-mini-instruct-FP8-KV

Phi-3.5-mini-instruct-FP8-KV

Model Overview

• Model Architecture: Phi-3.5
  • 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 , 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: 8/11/2024
• Version: 1.1
• License(s): mit
• Model Developers: Neural Magic

Quantized version of Phi-3.5-mini-instruct , with the new configuration files.

Model Optimizations

This model was obtained by quantizing the weights and activations of Phi-3.5-mini-instruct to FP8 data type, ready for inference with vLLM >= 0.5.1. 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/Phi-3.5-mini-instruct-FP8-KV"

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? Remember to respond in pirate speak!"},
]

prompts = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

llm = LLM(model=model_id, kv_cache_dtype="fp8")

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 LLM Compressor with calibration samples from UltraChat , as presented in the code snipet below.

from datasets import load_dataset
from transformers import AutoTokenizer

from llmcompressor.transformers import SparseAutoModelForCausalLM, oneshot

# Select model and load it.
# Phi-3.5 is a special case for KV cache quantization because it has
# fused QKV linear layers.
MODEL_ID = "microsoft/Phi-3.5-mini-instruct"
model = SparseAutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    device_map="auto",
    torch_dtype="auto",
)
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)

# Select calibration dataset.
DATASET_ID = "HuggingFaceH4/ultrachat_200k"
DATASET_SPLIT = "train_sft"

# Select number of samples. 512 samples is a good place to start.
# Increasing the number of samples can improve accuracy.
NUM_CALIBRATION_SAMPLES = 512
MAX_SEQUENCE_LENGTH = 2048

# Load dataset and preprocess.
ds = load_dataset(DATASET_ID, split=DATASET_SPLIT)
ds = ds.shuffle(seed=42).select(range(NUM_CALIBRATION_SAMPLES))


def process_and_tokenize(example):
    text = tokenizer.apply_chat_template(example["messages"], tokenize=False)
    return tokenizer(
        text,
        padding=False,
        max_length=MAX_SEQUENCE_LENGTH,
        truncation=True,
        add_special_tokens=False,
    )


ds = ds.map(process_and_tokenize, remove_columns=ds.column_names)

# Configure the quantization algorithm and scheme.
# In this case, we:
#   * quantize the weights to fp8 with per-tensor scales
#   * quantize the activations to fp8 with per-tensor scales
#   * quantize the kv cache to fp8 with per-tensor scales
recipe = """
quant_stage:
    quant_modifiers:
        QuantizationModifier:
            ignore: ["lm_head"]
            config_groups:
                group_0:
                    weights:
                        num_bits: 8
                        type: float
                        strategy: tensor
                        dynamic: false
                        symmetric: true
                    input_activations:
                        num_bits: 8
                        type: float
                        strategy: tensor
                        dynamic: false
                        symmetric: true
                    targets: ["Linear"]
            kv_cache_scheme:
                num_bits: 8
                type: float
                strategy: tensor
                dynamic: false
                symmetric: true
"""

# Apply algorithms.
oneshot(
    model=model,
    dataset=ds,
    recipe=recipe,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
)

# Confirm generations of the quantized model look sane.
print("\n\n")
print("========== SAMPLE GENERATION ==============")
input_ids = tokenizer("Hello my name is", return_tensors="pt").input_ids.to("cuda")
output = model.generate(input_ids, max_new_tokens=100)
print(tokenizer.decode(output[0]))
print("==========================================\n\n")

# Save to disk compressed.
SAVE_DIR = MODEL_ID.split("/")[1] + "-FP8-KV"
model.save_pretrained(SAVE_DIR, save_compressed=True)
tokenizer.save_pretrained(SAVE_DIR)

Evaluation

The model was evaluated on the OpenLLM leaderboard tasks (version 1) with the lm-evaluation-harness (commit 383bbd54bc621086e05aa1b030d8d4d5635b25e6) and the vLLM engine, using the following command:

lm_eval \
  --model vllm \
  --model_args pretrained="neuralmagic/Phi-3.5-mini-instruct-FP8-KV",kv_cache_dtype="fp8",gpu_memory_utilization=0.4,add_bos_token=True,max_model_len=4096 \
  --tasks openllm \
  --batch_size auto

Accuracy

Open LLM Leaderboard evaluation scores