Spaces:
Running
on
Zero
Running
on
Zero
Create model_utils.py
Browse files- model_utils.py +61 -0
model_utils.py
ADDED
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import subprocess
|
2 |
+
import os
|
3 |
+
import torch
|
4 |
+
from transformers import BitsAndBytesConfig, AutoConfig, AutoModelForCausalLM, LlavaNextForConditionalGeneration, LlavaForConditionalGeneration, PaliGemmaForConditionalGeneration, Idefics2ForConditionalGeneration
|
5 |
+
import spaces
|
6 |
+
|
7 |
+
os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
|
8 |
+
|
9 |
+
def install_flash_attn():
|
10 |
+
subprocess.run(
|
11 |
+
"pip install flash-attn --no-build-isolation",
|
12 |
+
env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
|
13 |
+
shell=True,
|
14 |
+
)
|
15 |
+
|
16 |
+
ARCHITECTURE_MAP = {
|
17 |
+
"LlavaNextForConditionalGeneration": LlavaNextForConditionalGeneration,
|
18 |
+
"LlavaForConditionalGeneration": LlavaForConditionalGeneration,
|
19 |
+
"PaliGemmaForConditionalGeneration": PaliGemmaForConditionalGeneration,
|
20 |
+
"Idefics2ForConditionalGeneration": Idefics2ForConditionalGeneration,
|
21 |
+
"AutoModelForCausalLM": AutoModelForCausalLM
|
22 |
+
}
|
23 |
+
|
24 |
+
@spaces.GPU
|
25 |
+
def get_model_summary(model_name):
|
26 |
+
try:
|
27 |
+
config = AutoConfig.from_pretrained(model_name)
|
28 |
+
architecture = config.architectures[0]
|
29 |
+
quantization_config = getattr(config, 'quantization_config', None)
|
30 |
+
|
31 |
+
if quantization_config:
|
32 |
+
bnb_config = BitsAndBytesConfig(
|
33 |
+
load_in_4bit=quantization_config.get('load_in_4bit', False),
|
34 |
+
load_in_8bit=quantization_config.get('load_in_8bit', False),
|
35 |
+
bnb_4bit_compute_dtype=quantization_config.get('bnb_4bit_compute_dtype', torch.float16),
|
36 |
+
bnb_4bit_quant_type=quantization_config.get('bnb_4bit_quant_type', 'nf4'),
|
37 |
+
bnb_4bit_use_double_quant=quantization_config.get('bnb_4bit_use_double_quant', False),
|
38 |
+
llm_int8_enable_fp32_cpu_offload=quantization_config.get('llm_int8_enable_fp32_cpu_offload', False),
|
39 |
+
llm_int8_has_fp16_weight=quantization_config.get('llm_int8_has_fp16_weight', False),
|
40 |
+
llm_int8_skip_modules=quantization_config.get('llm_int8_skip_modules', None),
|
41 |
+
llm_int8_threshold=quantization_config.get('llm_int8_threshold', 6.0),
|
42 |
+
)
|
43 |
+
else:
|
44 |
+
bnb_config = None
|
45 |
+
|
46 |
+
model_class = ARCHITECTURE_MAP.get(architecture, AutoModelForCausalLM)
|
47 |
+
model = model_class.from_pretrained(
|
48 |
+
model_name, config=bnb_config, trust_remote_code=True
|
49 |
+
)
|
50 |
+
|
51 |
+
if model and not quantization_config:
|
52 |
+
model = model.to(torch.device("cuda" if torch.cuda.is_available() else "cpu"))
|
53 |
+
|
54 |
+
model_summary = str(model) if model else "Model architecture not found."
|
55 |
+
return model_summary, ""
|
56 |
+
except ValueError as ve:
|
57 |
+
return "", f"ValueError: {ve}"
|
58 |
+
except EnvironmentError as ee:
|
59 |
+
return "", f"EnvironmentError: {ee}"
|
60 |
+
except Exception as e:
|
61 |
+
return "", str(e)
|