GLM-4.6-quantized.w4a16

GLM-4.6-quantized.w4a16

Model Overview

• Model Architecture: zai-org/GLM-4.6
  • Input: Text
  • Output: Text
• Model Optimizations:
  • Weight quantization: INT8
  • Activation quantization: INT8
• Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English.
• Version: 1.0
• Model Developers: RedHatAI

This model is a quantized version of zai-org/GLM-4.6. It was evaluated on a several tasks to assess the its quality in comparison to the unquatized model.

Model Optimizations

This model was obtained by quantizing the weights of zai-org/GLM-4.6 to INT4 data type, ready for inference with vLLM>=0.11.0.

Only the weights and activations of the linear operators within transformers blocks are quantized using LLM Compressor.

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 = "RedHatAI/GLM-4.6-quantized.w4a16"
number_gpus = 4

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, add_generation_prompt=True, tokenize=False)

llm = LLM(model=model_id, tensor_parallel_size=number_gpus)

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

<details>

from datasets import load_dataset
from transformers import AutoModelForCausalLM, AutoTokenizer

from llmcompressor import oneshot
from llmcompressor.modifiers.quantization import GPTQModifier
from llmcompressor.utils import dispatch_for_generation

MODEL_ID = "zai-org/GLM-4.6"

# Load model.
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID, 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.
# 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=f"{DATASET_SPLIT}[:{NUM_CALIBRATION_SAMPLES}]")
ds = ds.shuffle(seed=42)

def preprocess(example):
    return {
        "text": tokenizer.apply_chat_template(
            example["messages"],
            tokenize=False,
        )
    }

ds = ds.map(preprocess)

# Tokenize inputs.
def tokenize(sample):
    return tokenizer(
        sample["text"],
        padding=False,
        max_length=MAX_SEQUENCE_LENGTH,
        truncation=True,
        add_special_tokens=False,
    )

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

# Configure the quantization algorithm and scheme with explicit parameters.
recipe = GPTQModifier(
    targets="Linear",
    scheme="W4A16",
    ignore=[
        "lm_head",
        "re:.*mlp.gate$"
    ],
)

# Apply quantization.
oneshot(
    model=model,
    dataset=ds,
    recipe=recipe,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
    pipeline="sequential",
    sequential_targets=["Glm4MoeDecoderLayer"],  
    trust_remote_code_model=True,
)

SAVE_DIR = "./" + MODEL_ID.rstrip("/").split("/")[-1] + "-quantized.w4a16"
model.save_pretrained(SAVE_DIR, save_compressed=True)
tokenizer.save_pretrained(SAVE_DIR)

</details>

Evaluation

This model was evaluated on the well-known text benchmarks using lm-evaluation-harness. The Reasoning evals were done using ligheval.

Accuracy

<table> <thead> <tr> <th>Category</th> <th>Metric</th> <th>zai-org/GLM-4.6-FP8</th> <th>RedHatAI/GLM-4.6-quantized.w4a16 (this model)</th> <th>Recovery</th> </tr> </thead> <tbody> <!-- OpenLLM V1 --> <tr> <td rowspan="2"><b>Leaderboard</b></td> <td>MMLU Pro</td> <td>50.65%</td> <td>53.22%</td> <td>105.07%</td> </tr> <tr> <td>IFEVAL</td> <td>91.97%</td> <td>92.21%</td> <td>100.26%</td> </tr> <tr> <td rowspan="6"><b>Reasoning</b></td> <td>AIME25</td> <td>96.67%</td> <td>90.00%</td> <td>93.10%</td> </tr> <tr> <td>Math-500 (0-shot)</td> <td>88.80%</td> <td>88.00%</td> <td>99.10%</td> </tr> <tr> <td>GPQA (Diamond, 0-shot)</td> <td>81.82%</td> <td>80.30%</td> <td>98.14%</td> </tr> </tbody> </table>

Reproduction

The results were obtained using the following commands:

<details>

Leaderboard

lm_eval --model local-chat-completions \
  --tasks mmlu_pro  \
  --model_args "model=RedHatAI/GLM-4.6-quantized.w4a16,max_length=90000,base_url=http://0.0.0.0:3758/v1/chat/completions,num_concurrent=128,max_retries=3,tokenized_requests=False,tokenizer_backend=None,timeout=1200" \
  --num_fewshot 5 \
  --apply_chat_template \
  --fewshot_as_multiturn \
  --output_path ./ \
  --seed 42 \
  --gen_kwargs "do_sample=True,temperature=1.0,top_p=0.95,max_gen_toks=64000"


lm_eval --model local-chat-completions \
  --tasks leaderboard_ifeval  \
  --model_args "model=RedHatAI/GLM-4.6-quantized.w4a16,max_length=90000,base_url=http://0.0.0.0:3758/v1/chat/completions,num_concurrent=128,max_retries=3,tokenized_requests=False,tokenizer_backend=None,timeout=1200" \
  --num_fewshot 5 \
  --apply_chat_template \
  --fewshot_as_multiturn \
  --output_path ./ \
  --seed 42 \
  --gen_kwargs "do_sample=True,temperature=1.0,top_p=0.95,max_gen_toks=64000"

Reasoning

litellm_config.yaml:

model_parameters:
  provider: "hosted_vllm"
  model_name: "hosted_vllm/redhatai-glm-4.6-W4A16"
  base_url: "http://0.0.0.0:3759/v1"
  api_key: ""
  timeout: 3600
  concurrent_requests: 128
  generation_parameters:
    temperature: 1.0
    max_new_tokens: 131072
    top_p: 0.95
    seed: 0

lighteval endpoint litellm litellm_config.yaml \
  "aime25|0,math_500|0,gpqa:diamond|0" \
  --output-dir ./ \
  --save-details

</details>