Model Overview
่ฏฅๆจกๅ็จไบๅฏนAIๅๅค็ๆญฃ็กฎๆงๅๅ ๅฎนๅฏ็จไปทๅผ็่ฏไผฐ, ๅๆฐไป0-100. 0-35ๅฏไปฅๅคๆญไธบๅฎๅ จไธๅฏ็จ(ๅ ๅฎน้่ฏฏ,ไธไบๅฎไธ็ฌฆ,ๆช้ตๅพช็จๆท็ๆไปค). 35-50(้จๅๅ ๅฎนๅฏ็จ), 50-100(ๅ ๅฎนๆญฃ็กฎๅนถไธ้ตๅพช็จๆท็ๆไปค,ๅฏไปฅไฝฟ็จ). ่ฏฅๆจกๅไธ่ฌ้ ๅBest-of-Nๆฅไฝฟ็จ, ๅฏนไบ35ๅไปฅไธ็็ดๆฅไธขๅผ, ๅฏนไบ35-50ๅไน้ด็,ๅฏไปฅๅผๅ ฅcriticsๆฅไฟฎๆญฃๅ้ๆฐ้ๆ ท, 50ๅไปฅไธ็ๅฏไปฅ็ดๆฅไฝฟ็จ. ่ฏฅๆจกๅๆขๅฏไปฅ็จไบFinal Answer็่ฏไผฐ, ไนๅฏไปฅ็จไบLLM่ฐ็จTool็่ฏไผฐ(ไธป่ฆ่ฏไผฐไฝฟ็จๅทฅๅ ทๆฏๅฆๅ็,ๅๆฐๆฏๅฆๆญฃ็กฎ.)
SCORE: 0.00-100.00
Usage
Run the Inference Code
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
model_id = "weiminw/Heliumos-RM-4B"
tokenizer = AutoTokenizer.from_pretrained(model_id,padding_side="left")
model = AutoModelForSequenceClassification.from_pretrained(
model_id,
torch_dtype="auto",
device_map="auto"
)
messages = [
{'role': 'user', 'content': "what is 12 * 12?"},
{'role': 'assistant', 'content': "144"}
]
text_encoded = tokenizer.apply_chat_template(messages, return_dict=True, return_tensors="pt", tokenize=True).to("cuda:0")
score = model(**text_encoded)
print(score.logits[0]) #
- Downloads last month
- 14