File size: 4,457 Bytes
d4b224c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
---
base_model:
- Qwen/Qwen3-VL-30B-A3B-Instruct
---

How to quant:

git clone https://github.com/vllm-project/llm-compressor.git

cd llm-compressor

pip install -e .

pip install transformers==4.57.0

export TOKENIZERS_PARALLELISM=false

cd ..

Script For quant:

Save the next code like quant.py.

python3 quant.py

```
import torch
from datasets import load_dataset
from transformers import AutoProcessor, Qwen3VLMoeForConditionalGeneration

from llmcompressor import oneshot
from llmcompressor.modeling import replace_modules_for_calibration
from llmcompressor.modifiers.awq import AWQModifier
from llmcompressor.utils import dispatch_for_generation

# NOTE: Requires a minimum of transformers 4.57.0

MODEL_ID = "Qwen/Qwen3-VL-30B-A3B-Instruct"

# Load model.
model = Qwen3VLMoeForConditionalGeneration.from_pretrained(
    MODEL_ID, 
    torch_dtype=torch.bfloat16,
    device_map=None,
    trust_remote_code=True
)
processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
model = replace_modules_for_calibration(model)

DATASET_ID = "neuralmagic/calibration"
NUM_CALIBRATION_SAMPLES = 256
MAX_SEQUENCE_LENGTH = 8192

ds = load_dataset(DATASET_ID, name="LLM", split=f"train[:{NUM_CALIBRATION_SAMPLES}]")
ds = ds.shuffle(seed=42)


def preprocess_function(example):
    messages = []
    for message in example["messages"]:
        messages.append(
            {
                "role": message["role"],
                "content": [{"type": "text", "text": message["content"]}],
            }
        )

    return processor.apply_chat_template(
        messages,
        return_tensors="pt",
        padding=False,
        truncation=True,
        max_length=MAX_SEQUENCE_LENGTH,
        tokenize=True,
        add_special_tokens=False,
        return_dict=True,
        add_generation_prompt=False,
    )


ds = ds.map(preprocess_function, batched=False, remove_columns=ds.column_names)


def data_collator(batch):
    assert len(batch) == 1
    return {
        key: (
            torch.tensor(value)
            if key != "pixel_values"
            else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
        )
        for key, value in batch[0].items()
    }


# Configure AWQ quantization with smoothing and balancing
recipe = AWQModifier(
    ignore=[
        're:.*embed_tokens', 
        're:.*input_layernorm$', 
        're:.*mlp[.]gate$', 
        're:.*post_attention_layernorm$', 
        're:.*norm$', 
        're:model[.]visual.*',
        're:visual.*',
        'lm_head'
    ],
    mappings=[
        {
            "smooth_layer": "re:.*input_layernorm$",
            "balance_layers": ['re:.*q_proj$', 're:.*k_proj$', 're:.*v_proj$']
        },
        {
            "smooth_layer": "re:.*v_proj$",
            "balance_layers": ['re:.*o_proj$']
        },
        {
            "smooth_layer": "re:.*post_attention_layernorm$",
            "balance_layers": ['re:.*gate_proj$', 're:.*up_proj$']
        },
        {
            "smooth_layer": "re:.*up_proj$",
            "balance_layers": ['re:.*down_proj$']
        }
    ],
    duo_scaling=True,
    config_groups={
        "group_0": {
            "targets": ["Linear"],
            "weights": {
                "num_bits": 4,
                "type": "int",
                "symmetric": True,
                "group_size": 32,
                "strategy": "group",
                "block_structure": None,
                "dynamic": False,
                "actorder": None,
                "observer": "mse",
                "observer_kwargs": {}
            },
            "input_activations": None,
            "output_activations": None,
            "format": None
        }
    }
)

# Apply AWQ quantization.
oneshot(
    model=model,
    processor=processor,
    recipe=recipe,
    dataset=ds,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
    data_collator=data_collator,

)

print("========== SAMPLE GENERATION ==============")
dispatch_for_generation(model)
input_ids = processor(text="Hello my name is", return_tensors="pt").input_ids.to("cuda")
output = model.generate(input_ids, max_new_tokens=20)
print(processor.decode(output[0]))
print("==========================================")

# Save to disk in compressed-tensors format.
SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-AWQ-W4A16-mse-seq"
model.save_pretrained(SAVE_DIR, save_compressed=True)
processor.save_pretrained(SAVE_DIR)
```