LiqiangJing
commented
1 year ago disable_torch_init()
load_bf16 = True
model_path = "/home/xxx/LLaVA-RLHF/llava/eval/checkpoints/sft_model"
lora_path = "/home/xxx/LLaVA-RLHF/llava/eval/checkpoints/rlhf_lora_adapter_model"
model_name = "LLaVA-RLHF-13b-v1.5-336"
tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name,
load_bf16=load_bf16)
model = PeftModel.from_pretrained(
model,
lora_path,
)
questions = [json.loads(q) for q in open(os.path.expanduser(args.question_file), "r")] ###
questions = get_chunk(questions, args.num_chunks, args.chunk_idx)
answers_file = os.path.expanduser(args.answers_file)
os.makedirs(os.path.dirname(answers_file), exist_ok=True)
ans_file = open(answers_file, "w")
for line in tqdm(questions):
idx = line["id"]
image_file = line["image"]
qs = line["instruction"]
cur_prompt = qs
if model.config.mm_use_im_start_end:
qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs
else:
qs = DEFAULT_IMAGE_TOKEN + '\n' + qs
conv = conv_templates[args.conv_mode].copy()
conv.append_message(conv.roles[0], qs)
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda()
image = Image.open(os.path.join(args.image_folder, image_file))
image_tensor = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0].to(dtype=torch.bfloat16)
stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
keywords = [stop_str]
stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids)
model = model.merge_and_unload()
with torch.inference_mode():
output_ids = model.generate(
input_ids,
images=image_tensor.unsqueeze(0).half().cuda(),
do_sample=True,
temperature=args.temperature,
top_p=args.top_p,
num_beams=args.num_beams,
# no_repeat_ngram_size=3,
max_new_tokens=1024,
use_cache=True)
input_token_len = input_ids.shape[1]
n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item()
if n_diff_input_output > 0:
print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids')
outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0]
outputs = outputs.strip()
if outputs.endswith(stop_str):
outputs = outputs[:-len(stop_str)]
outputs = outputs.strip()
ans_id = shortuuid.uuid()
ans_file.write(json.dumps({"question_id": idx,
"prompt": cur_prompt,
"text": outputs,
"answer_id": ans_id,
"model_id": model_name,
"metadata": {}}) + "\n")
ans_file.flush()
ans_file.close()
Edward-Sun
yuvalkirstain
I used the model_vqa.py but I met the mistake for LLaVA-RLHF