| from PIL import Image |
| import torch |
|
|
| from .base import BaseModel |
| from ..smp import * |
|
|
|
|
| class PaliGemma(BaseModel): |
| INSTALL_REQ = False |
| INTERLEAVE = False |
|
|
| def __init__(self, model_path='google/paligemma-3b-mix-448', **kwargs): |
| try: |
| from transformers import AutoProcessor, PaliGemmaForConditionalGeneration |
| except Exception as e: |
| logging.critical('Please install the latest version transformers.') |
| raise e |
|
|
| model = PaliGemmaForConditionalGeneration.from_pretrained( |
| model_path, |
| torch_dtype=torch.bfloat16, |
| device_map='cpu', |
| revision='bfloat16', |
| ).eval() |
| self.model = model.cuda() |
| self.processor = AutoProcessor.from_pretrained(model_path) |
| self.kwargs = kwargs |
|
|
| def generate_inner(self, message, dataset=None): |
| prompt, image_path = self.message_to_promptimg(message, dataset=dataset) |
| image = Image.open(image_path).convert('RGB') |
|
|
| model_inputs = self.processor( |
| text=prompt, images=image, return_tensors='pt' |
| ).to('cuda') |
| input_len = model_inputs['input_ids'].shape[-1] |
|
|
| with torch.inference_mode(): |
| generation = self.model.generate( |
| **model_inputs, max_new_tokens=512, do_sample=False |
| ) |
| generation = generation[0][input_len:] |
| res = self.processor.decode(generation, skip_special_tokens=True) |
| return res |
|
|
|
|
| class Gemma3(BaseModel): |
|
|
| INSTALL_REQ = False |
| INTERLEAVE = True |
|
|
| def __init__(self, model_path='google/gemma-3-4b-it', **kwargs): |
| logging.info( |
| "Please install transformers via \n" |
| "pip install git+https://github.com/huggingface/transformers@v4.49.0-Gemma-3" |
| ) |
| try: |
| from transformers import AutoProcessor, Gemma3ForConditionalGeneration |
| import torch |
| except Exception as e: |
| logging.critical('Please install torch and transformers') |
| raise e |
|
|
| self.model = Gemma3ForConditionalGeneration.from_pretrained( |
| model_path, device_map="cuda", attn_implementation="flash_attention_2" |
| ).eval() |
|
|
| self.device = self.model.device |
| self.processor = AutoProcessor.from_pretrained(model_path) |
|
|
| self.system_prompt = kwargs.pop('system_prompt', 'You are a helpful assistant. ') |
|
|
| default_kwargs = { |
| 'do_sample': False, |
| 'max_new_tokens': 2048 |
| } |
| default_kwargs.update(kwargs) |
| self.kwargs = default_kwargs |
|
|
| def message2pipeline(self, message): |
| ret = [] |
| if hasattr(self, 'system_prompt') and self.system_prompt is not None: |
| ret = [ |
| dict(role='system', content=[dict(type='text', text=self.system_prompt)]) |
| ] |
| content = [] |
| for m in message: |
| if m['type'] == 'text': |
| content.append(dict(type='text', text=m['value'])) |
| elif m['type'] == 'image': |
| content.append(dict(type='image', url=m['value'])) |
| ret.append(dict(role='user', content=content)) |
| return ret |
|
|
| def generate_inner(self, message, dataset=None): |
| messages = self.message2pipeline(message) |
| inputs = self.processor.apply_chat_template( |
| messages, add_generation_prompt=True, tokenize=True, |
| return_dict=True, return_tensors="pt", |
| ).to(self.device, dtype=torch.bfloat16) |
|
|
| input_len = inputs['input_ids'].shape[-1] |
| |
| with torch.inference_mode(): |
| generation = self.model.generate(**inputs, **self.kwargs) |
| generation = generation[0][input_len:] |
|
|
| decoded = self.processor.decode(generation, skip_special_tokens=True) |
| return decoded |
|
|
