metadata
library_name: transformers
tags:
- unsloth
- trl
- sft
datasets:
- mintaeng/llm_futsaldata_yo
license: apache-2.0
language:
- ko
FUT FUT CHAT BOT
- μ€νμμ€ λͺ¨λΈμ LLM fine tuning κ³Ό RAG λ₯Ό μ μ©ν μμ±ν AI
- νμ΄μ λν κ΄μ¬μ΄ λμμ§λ©΄μ μμ λλΉ μ λ¬Έμλ₯Ό μν μ 보 μ 곡 μλΉμ€κ° νμνλ€κ³ λκ»΄ μ μνκ² λ¨
- νμ΄ νλ«νΌμ μ¬μ©λλ νμ΄ μ 보 λμ°λ―Έ μ±λ΄
- 'ν΄μ체'λ‘ λ΅νλ©° λ¬Έμ₯ λμ 'μΌλ§λ μ§ λ¬Όμ΄λ³΄μΈμ~ νν~!' μ μΆλ ₯ν¨
- train for 7h23m
HOW TO USE
#!pip install transformers==4.40.0 accelerate
import os
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
model_id = 'Dongwookss/small_fut_final'
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.bfloat16,
device_map="auto",
)
model.eval()
Query
from transformers import TextStreamer
PROMPT = '''Below is an instruction that describes a task. Write a response that appropriately completes the request.
μ μνλ contextμμλ§ λλ΅νκ³ contextμ μλ λ΄μ©μ λͺ¨λ₯΄κ² λ€κ³ λλ΅ν΄'''
messages = [
{"role": "system", "content": f"{PROMPT}"},
{"role": "user", "content": f"{instruction}"}
]
input_ids = tokenizer.apply_chat_template(
messages,
add_generation_prompt=True,
return_tensors="pt"
).to(model.device)
terminators = [
tokenizer.eos_token_id,
tokenizer.convert_tokens_to_ids("<|eot_id|>")
]
text_streamer = TextStreamer(tokenizer)
_ = model.generate(
input_ids,
max_new_tokens=4096,
eos_token_id=terminators,
do_sample=True,
streamer = text_streamer,
temperature=0.6,
top_p=0.9,
repetition_penalty = 1.1
)
Model Details
Model Description
This is the model card of a π€ transformers model that has been pushed on the Hub. This model card has been automatically generated.
- Developed by: Dongwookss
- Model type: text generation
- Language(s) (NLP): Korean
- Finetuned from model : HuggingFaceH4/zephyr-7b-beta
Data
https://huggingface.co/datasets/mintaeng/llm_futsaldata_yo
νμ΅ λ°μ΄ν°μ μ beomi/KoAlpaca-v1.1a λ₯Ό λ² μ΄μ€λ‘ μΆκ°, ꡬμΆ, μ μ²λ¦¬ μ§νν 23.5k λ°μ΄ν°λ‘ νλνμμ΅λλ€. λ°μ΄ν°μ μ instruction, input, output μΌλ‘ ꡬμ±λμ΄ μμΌλ©° tuning λͺ©νμ λ§κ² λ§ν¬ μμ νμμ΅λλ€. λλ©μΈ μ 보μ λν λ°μ΄ν° μΆκ°νμμ΅λλ€.
Training & Result
Training Procedure
LoRAμ SFT Trainer λ°©μμ μ¬μ©νμμ΅λλ€.
Training Hyperparameters
- Training regime: bf16 mixed precision
r=32,
lora_alpha=64, # QLoRA : alpha = r/2 // LoRA : alpha =r*2
lora_dropout=0.05,
target_modules=[
"q_proj",
"k_proj",
"v_proj",
"o_proj",
"gate_proj",
"up_proj",
"down_proj",
], # νκ² λͺ¨λ
Result
https://github.com/lucide99/Chatbot_FutFut
Environment
L4 GPU