EEVE-Korean-Custom-10.8B
๐ฐ๐ท Korean Custom Fine-tuning - Responds politely in formal Korean even to casual questions
English Documentation
Model Overview
This model is based on EEVE-Korean-Instruct-10.8B-v1.0, fine-tuned with high-quality Korean instruction data using LoRA, and subsequently merged into a standalone model.
Key Features:
- High-quality Korean language processing trained on 100K+ instruction samples
- Extended context support up to 8K tokens
- Bilingual capabilities supporting both Korean and English
Quick Start
Installation:
pip install transformers torch accelerate
Basic Usage:
from transformers import AutoModelForCausalLM, AutoTokenizer
# Load model (no PEFT required)
model = AutoModelForCausalLM.from_pretrained(
"MyeongHo0621/eeve-vss-smh",
device_map="auto",
torch_dtype="auto"
)
tokenizer = AutoTokenizer.from_pretrained("MyeongHo0621/eeve-vss-smh")
# Prompt template (EEVE format)
def create_prompt(user_input):
return f"""A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
Human: {user_input}
Assistant: """
# Generate response
user_input = "Implement Fibonacci sequence in Python"
prompt = create_prompt(user_input)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(
**inputs,
max_new_tokens=512,
temperature=0.3,
top_p=0.85,
repetition_penalty=1.0,
do_sample=True
)
response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
print(response)
Streaming Generation:
from transformers import TextIteratorStreamer
from threading import Thread
streamer = TextIteratorStreamer(tokenizer, skip_special_tokens=True)
generation_kwargs = {
**inputs,
"max_new_tokens": 512,
"temperature": 0.3,
"top_p": 0.85,
"streamer": streamer
}
thread = Thread(target=model.generate, kwargs=generation_kwargs)
thread.start()
for text in streamer:
print(text, end="", flush=True)
Training Details
Dataset Configuration:
- Size: Approximately 100,000 samples
- Sources: Combination of high-quality Korean instruction datasets including KoAlpaca, Ko-Ultrachat, KoInstruct, Kullm-v2, Smol Korean Talk, and Korean Wiki QA
- Preprocessing: Length filtering, deduplication, language verification, and special character removal
LoRA Configuration:
r: 128 # Higher rank for stronger learning
lora_alpha: 256 # alpha = 2 * r
lora_dropout: 0.0 # No dropout (Unsloth optimization)
target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
- gate_proj
- up_proj
- down_proj
bias: none
task_type: CAUSAL_LM
use_rslora: false
Training Hyperparameters:
| Parameter | Value | Description |
|---|---|---|
| Framework | Unsloth | 2-5x faster than standard transformers |
| Epochs | 3 (stopped at 1.94) | Early stopping at optimal point |
| Batch Size | 8 per device | Maximizing H100E memory |
| Gradient Accumulation | 2 | Effective batch size of 16 |
| Learning Rate | 1e-4 | Balanced learning rate |
| Max Sequence Length | 4096 | Extended context support |
| Warmup Ratio | 0.05 | Quick warmup |
| Weight Decay | 0.01 | Regularization |
| Optimizer | AdamW 8-bit (Unsloth) | Memory optimized |
| LR Scheduler | Cosine | Smooth decay |
| Gradient Checkpointing | Unsloth optimized | Memory efficient |
Checkpoint Selection Strategy:
The model was trained for 3 epochs, but we selected checkpoint-6250 (Epoch 1.94) based on evaluation loss analysis:
| Checkpoint | Epoch | Training Loss | Eval Loss | Status |
|---|---|---|---|---|
| 6250 | 1.94 | 0.9986 | 1.4604 | โ Selected (Best) |
| 6500 | 2.02 | 0.561 | 1.5866 | โ Overfitting |
Key Insight: Training loss continued to decrease, but evaluation loss started increasing after checkpoint-6250, indicating overfitting. We selected the checkpoint with the lowest evaluation loss for optimal generalization.
Memory Optimization:
- Full precision training (no 4-bit quantization needed on H100E)
- Unsloth gradient checkpointing
- BF16 training optimized for H100E
- Peak VRAM usage: ~26GB during training
Training Infrastructure:
- GPU: NVIDIA H100 80GB HBM3
- Framework: Unsloth + PyTorch 2.6, Transformers 4.46.3
- Training time: ~3 hours (6,250 steps with Unsloth acceleration)
- Final checkpoint: Step 6250 (Epoch 1.94), merged to full model
Performance Examples
Casual to Formal Korean Conversion:
Input (casual Korean): "WMS๊ฐ ๋ญ์ผ?"
Output (formal Korean): "WMS๋ Warehouse Management System์ ์ฝ์๋ก, ์ฐฝ๊ณ ๊ด๋ฆฌ ์์คํ ์ ์๋ฏธํฉ๋๋ค. ์ฌ๊ณ ์ถ์ , ์ ์ถ๊ณ ๊ด๋ฆฌ, ํผํน, ํจํน ๋ฑ์ ๋ฌผ๋ฅ ํ๋ก์ธ์ค๋ฅผ ์๋ํํ๊ณ ์ต์ ํํ๋ ์ํํธ์จ์ด ์์คํ ์ ๋๋ค. ํจ์จ์ ์ธ ์ฐฝ๊ณ ์ด์์ ์ํด ์ฌ์ฉ๋๋ฉฐ, ์ค์๊ฐ ์ฌ๊ณ ๊ฐ์์ฑ๊ณผ ์์ ์์ฐ์ฑ ํฅ์์ ์ ๊ณตํฉ๋๋ค."
Code Generation:
Input: "ํ์ด์ฌ์ผ๋ก ๋ฆฌ์คํธ๋ฅผ ์ญ์์ผ๋ก ๋ง๋ค์ด์ค"
Output: Provides three different Python methods for list reversal with detailed explanations of each approach, including reverse() method, slicing, and reversed() function, along with their differences.
Prompt Template
This model uses the standard EEVE template format:
template = """A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
Human: {user_message}
Assistant: """
Using this exact template is essential for optimal performance.
Recommended Generation Parameters
generation_config = {
"max_new_tokens": 512,
"temperature": 0.3,
"top_p": 0.85,
"repetition_penalty": 1.0,
"do_sample": True,
"pad_token_id": tokenizer.pad_token_id,
"eos_token_id": tokenizer.eos_token_id,
}
Parameter Tuning Guide:
| Use Case | Temperature | Top P | Repetition Penalty | Notes |
|---|---|---|---|---|
| Precise answers | 0.1-0.3 | 0.8-0.9 | 1.0 | Best for factual Q&A |
| Balanced responses | 0.5-0.7 | 0.85-0.95 | 1.0 | Recommended default |
| Creative outputs | 0.8-1.0 | 0.9-1.0 | 1.05-1.1 | For creative writing |
Important Notes on Repetition Penalty:
- Default (1.0): No penalty, natural repetition allowed
- Light (1.05-1.1): Reduces minor repetition in creative tasks
- Moderate (1.1-1.2): Good for reducing repetitive phrases
- Strong (1.2+): May affect output quality, use with caution
โ ๏ธ Warning: Setting repetition_penalty > 1.2 can degrade Korean text quality. For this model, 1.0-1.1 is optimal for most use cases.
Advanced Configuration Example:
# For code generation
code_gen_config = {
"max_new_tokens": 1024,
"temperature": 0.2,
"top_p": 0.9,
"repetition_penalty": 1.0,
"do_sample": True,
}
# For conversational responses
conversation_config = {
"max_new_tokens": 512,
"temperature": 0.7,
"top_p": 0.9,
"repetition_penalty": 1.05,
"do_sample": True,
}
# For precise factual answers
factual_config = {
"max_new_tokens": 256,
"temperature": 0.1,
"top_p": 0.85,
"repetition_penalty": 1.0,
"do_sample": True,
}
Limitations
This model has been released for research and educational purposes. Commercial use requires compliance with the CC-BY-NC-SA-4.0 license. While optimized for Korean language, the model provides partial support for other languages. Performance may improve with additional training beyond checkpoint 500.
License
- Model License: CC-BY-NC-SA-4.0
- Base Model: Complies with EEVE-Korean-Instruct-10.8B-v1.0 license
- Commercial Use: Restricted (refer to license)
Citation
@misc{eeve-vss-smh-2025,
author = {MyeongHo0621},
title = {EEVE-VSS-SMH: Korean Custom Fine-tuned Model},
year = {2025},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/MyeongHo0621/eeve-vss-smh}},
note = {LoRA fine-tuned and merged model based on EEVE-Korean-Instruct-10.8B-v1.0}
}
Acknowledgments
- Base Model: Yanolja - EEVE-Korean-Instruct-10.8B-v1.0
- Training Infrastructure: KT Cloud H100E
- Framework: Hugging Face Transformers, PEFT
Contact
- GitHub: MyeongHo0621
- Model Repository: tesseract
ํ๊ตญ์ด ๋ฌธ์
๋ชจ๋ธ ์๊ฐ
์ด ๋ชจ๋ธ์ EEVE-Korean-Instruct-10.8B-v1.0์ ๋ฒ ์ด์ค๋ก, ๊ณ ํ์ง ํ๊ตญ์ด instruction ๋ฐ์ดํฐ๋ก LoRA ํ์ธํ๋ํ ํ ๋ณํฉํ ๋ชจ๋ธ์ ๋๋ค.
์ฃผ์ ํน์ง:
- 100K+ ๊ณ ํ์ง instruction ๋ฐ์ดํฐ๋ก ํ๋ จ๋ ํ๊ตญ์ด ์ฒ๋ฆฌ ๋ฅ๋ ฅ
- ์ต๋ 8K ํ ํฐ๊น์ง ํ์ฅ๋ ๋ฌธ๋งฅ ์ง์
- ํ๊ตญ์ด์ ์์ด๋ฅผ ๋ชจ๋ ์ง์ํ๋ ์ด์ค์ธ์ด ๊ธฐ๋ฅ
๋น ๋ฅธ ์์
์ค์น:
pip install transformers torch accelerate
๊ธฐ๋ณธ ์ฌ์ฉ:
from transformers import AutoModelForCausalLM, AutoTokenizer
# ๋ชจ๋ธ ๋ก๋ (PEFT ๋ถํ์)
model = AutoModelForCausalLM.from_pretrained(
"MyeongHo0621/eeve-vss-smh",
device_map="auto",
torch_dtype="auto"
)
tokenizer = AutoTokenizer.from_pretrained("MyeongHo0621/eeve-vss-smh")
# ํ๋กฌํํธ ํ
ํ๋ฆฟ (EEVE ํ์)
def create_prompt(user_input):
return f"""A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
Human: {user_input}
Assistant: """
# ์๋ต ์์ฑ
user_input = "ํ์ด์ฌ์ผ๋ก ํผ๋ณด๋์น ์์ด ๊ตฌํํด์ค"
prompt = create_prompt(user_input)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(
**inputs,
max_new_tokens=512,
temperature=0.3,
top_p=0.85,
repetition_penalty=1.0,
do_sample=True
)
response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
print(response)
์คํธ๋ฆฌ๋ฐ ์์ฑ:
from transformers import TextIteratorStreamer
from threading import Thread
streamer = TextIteratorStreamer(tokenizer, skip_special_tokens=True)
generation_kwargs = {
**inputs,
"max_new_tokens": 512,
"temperature": 0.3,
"top_p": 0.85,
"streamer": streamer
}
thread = Thread(target=model.generate, kwargs=generation_kwargs)
thread.start()
for text in streamer:
print(text, end="", flush=True)
ํ๋ จ ์ธ๋ถ์ฌํญ
๋ฐ์ดํฐ์ ๊ตฌ์ฑ:
- ํฌ๊ธฐ: ์ฝ 100,000๊ฐ ์ํ
- ์ถ์ฒ: KoAlpaca, Ko-Ultrachat, KoInstruct, Kullm-v2, Smol Korean Talk, Korean Wiki QA ๋ฑ ๊ณ ํ์ง ํ๊ตญ์ด instruction ๋ฐ์ดํฐ์ ์กฐํฉ
- ์ ์ฒ๋ฆฌ: ๊ธธ์ด ํํฐ๋ง, ์ค๋ณต ์ ๊ฑฐ, ์ธ์ด ํ์ธ, ํน์๋ฌธ์ ์ ๊ฑฐ
LoRA ์ค์ :
r: 128 # ๋ ๋์ rank (๊ฐ๋ ฅํ ํ์ต)
lora_alpha: 256 # alpha = 2 * r
lora_dropout: 0.0 # Dropout ์์ (Unsloth ์ต์ ํ)
target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
- gate_proj
- up_proj
- down_proj
bias: none
task_type: CAUSAL_LM
use_rslora: false
ํ๋ จ ํ์ดํผํ๋ผ๋ฏธํฐ:
| ํ๋ผ๋ฏธํฐ | ๊ฐ | ์ค๋ช |
|---|---|---|
| ํ๋ ์์ํฌ | Unsloth | ๊ธฐ์กด ๋๋น 2-5๋ฐฐ ๋น ๋ฅธ ํ๋ จ |
| Epochs | 3 (1.94์์ ์ค๋จ) | ์ต์ ์ง์ ์์ ์กฐ๊ธฐ ์ข ๋ฃ |
| Batch Size | 8 per device | H100E ๋ฉ๋ชจ๋ฆฌ ์ต๋ ํ์ฉ |
| Gradient Accumulation | 2 | ์ค์ง์ ๋ฐฐ์น ํฌ๊ธฐ 16 |
| Learning Rate | 1e-4 | ๊ท ํ์กํ ํ์ต๋ฅ |
| Max Sequence Length | 4096 | ํ์ฅ๋ ๋ฌธ๋งฅ ์ง์ |
| Warmup Ratio | 0.05 | ๋น ๋ฅธ ์๋ฐ์ |
| Weight Decay | 0.01 | ์ ๊ทํ |
| Optimizer | AdamW 8-bit (Unsloth) | ๋ฉ๋ชจ๋ฆฌ ์ต์ ํ |
| LR Scheduler | Cosine | ๋ถ๋๋ฌ์ด ๊ฐ์ |
| Gradient Checkpointing | Unsloth ์ต์ ํ | ๋ฉ๋ชจ๋ฆฌ ํจ์จ |
์ฒดํฌํฌ์ธํธ ์ ํ ์ ๋ต:
3 epoch ํ๋ จ์ ์งํํ์ง๋ง, ํ๊ฐ ์์ค(evaluation loss) ๋ถ์์ ํตํด **checkpoint-6250 (Epoch 1.94)**์ ์ ํํ์ต๋๋ค:
| ์ฒดํฌํฌ์ธํธ | Epoch | Training Loss | Eval Loss | ์ํ |
|---|---|---|---|---|
| 6250 | 1.94 | 0.9986 | 1.4604 | โ ์ ํ (์ต์ ) |
| 6500 | 2.02 | 0.561 | 1.5866 | โ ๊ณผ์ ํฉ |
ํต์ฌ ์ธ์ฌ์ดํธ: Training loss๋ ๊ณ์ ๊ฐ์ํ์ง๋ง, checkpoint-6250 ์ดํ evaluation loss๊ฐ ์ฆ๊ฐํ๊ธฐ ์์ํ์ต๋๋ค. ์ด๋ ๊ณผ์ ํฉ์ ์ ํธ์ ๋๋ค. ๊ฐ์ฅ ๋ฎ์ evaluation loss๋ฅผ ๊ฐ์ง ์ฒดํฌํฌ์ธํธ๋ฅผ ์ ํํ์ฌ ์ต์ ์ ์ผ๋ฐํ ์ฑ๋ฅ์ ํ๋ณดํ์ต๋๋ค.
๋ฉ๋ชจ๋ฆฌ ์ต์ ํ:
- Full precision ํ๋ จ (H100E์์ 4-bit ์์ํ ๋ถํ์)
- Unsloth gradient checkpointing
- H100E ์ต์ ํ BF16 ํ๋ จ
- ํ๋ จ ์ค Peak VRAM ์ฌ์ฉ๋: ~26GB
ํ๋ จ ํ๊ฒฝ:
- GPU: NVIDIA H100 80GB HBM3
- ํ๋ ์์ํฌ: Unsloth + PyTorch 2.6, Transformers 4.46.3
- ํ๋ จ ์๊ฐ: ~3์๊ฐ (Unsloth ๊ฐ์์ผ๋ก 6,250 steps)
- ์ต์ข ์ฒดํฌํฌ์ธํธ: Step 6250 (Epoch 1.94), ์ ์ฒด ๋ชจ๋ธ๋ก ๋ณํฉ
์ฑ๋ฅ ์์
๋ฐ๋ง์์ ์กด๋๋ง ๋ณํ:
์ ๋ ฅ (๋ฐ๋ง): "WMS๊ฐ ๋ญ์ผ?"
์ถ๋ ฅ (์กด๋๋ง): "WMS๋ Warehouse Management System์ ์ฝ์๋ก, ์ฐฝ๊ณ ๊ด๋ฆฌ ์์คํ ์ ์๋ฏธํฉ๋๋ค. ์ฌ๊ณ ์ถ์ , ์ ์ถ๊ณ ๊ด๋ฆฌ, ํผํน, ํจํน ๋ฑ์ ๋ฌผ๋ฅ ํ๋ก์ธ์ค๋ฅผ ์๋ํํ๊ณ ์ต์ ํํ๋ ์ํํธ์จ์ด ์์คํ ์ ๋๋ค. ํจ์จ์ ์ธ ์ฐฝ๊ณ ์ด์์ ์ํด ์ฌ์ฉ๋๋ฉฐ, ์ค์๊ฐ ์ฌ๊ณ ๊ฐ์์ฑ๊ณผ ์์ ์์ฐ์ฑ ํฅ์์ ์ ๊ณตํฉ๋๋ค."
์ฝ๋ ์์ฑ:
์ ๋ ฅ: "ํ์ด์ฌ์ผ๋ก ๋ฆฌ์คํธ๋ฅผ ์ญ์์ผ๋ก ๋ง๋ค์ด์ค"
์ถ๋ ฅ: reverse() ๋ฉ์๋, ์ฌ๋ผ์ด์ฑ, reversed() ํจ์ ๋ฑ ์ธ ๊ฐ์ง ํ์ด์ฌ ๋ฆฌ์คํธ ์ญ์ ๋ณํ ๋ฐฉ๋ฒ์ ๊ฐ ์ ๊ทผ๋ฒ์ ์ฐจ์ด์ ๊ณผ ํจ๊ป ์์ธํ ์ค๋ช ํฉ๋๋ค.
ํ๋กฌํํธ ํ ํ๋ฆฟ
์ด ๋ชจ๋ธ์ ํ์ค EEVE ํ ํ๋ฆฟ ํ์์ ์ฌ์ฉํฉ๋๋ค:
template = """A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
Human: {user_message}
Assistant: """
์ต์ ์ ์ฑ๋ฅ์ ์ํด์๋ ์ด ํ ํ๋ฆฟ์ ์ ํํ ์ฌ์ฉํ๋ ๊ฒ์ด ํ์์ ์ ๋๋ค.
๊ถ์ฅ ์์ฑ ํ๋ผ๋ฏธํฐ
generation_config = {
"max_new_tokens": 512,
"temperature": 0.3,
"top_p": 0.85,
"repetition_penalty": 1.0,
"do_sample": True,
"pad_token_id": tokenizer.pad_token_id,
"eos_token_id": tokenizer.eos_token_id,
}
ํ๋ผ๋ฏธํฐ ์กฐ์ ๊ฐ์ด๋:
| ์ฉ๋ | Temperature | Top P | Repetition Penalty | ๋น๊ณ |
|---|---|---|---|---|
| ์ ํํ ๋ต๋ณ | 0.1-0.3 | 0.8-0.9 | 1.0 | ์ฌ์ค ๊ธฐ๋ฐ Q&A์ ์ต์ |
| ๊ท ํ์กํ ๋ต๋ณ | 0.5-0.7 | 0.85-0.95 | 1.0 | ๊ถ์ฅ ๊ธฐ๋ณธ๊ฐ |
| ์ฐฝ์์ ๋ต๋ณ | 0.8-1.0 | 0.9-1.0 | 1.05-1.1 | ์ฐฝ์ ๊ธ์ฐ๊ธฐ์ฉ |
Repetition Penalty ์ค์ ์ฐธ๊ณ ์ฌํญ:
- ๊ธฐ๋ณธ๊ฐ (1.0): ํ๋ํฐ ์์, ์์ฐ์ค๋ฌ์ด ๋ฐ๋ณต ํ์ฉ
- ์ฝํจ (1.05-1.1): ์ฐฝ์ ์์ ์์ ๋ฏธ์ธํ ๋ฐ๋ณต ๊ฐ์
- ์ค๊ฐ (1.1-1.2): ๋ฐ๋ณต์ ์ธ ๊ตฌ๋ฌธ ๊ฐ์์ ํจ๊ณผ์
- ๊ฐํจ (1.2+): ์ถ๋ ฅ ํ์ง ์ ํ ๊ฐ๋ฅ, ์ฃผ์ํด์ ์ฌ์ฉ
โ ๏ธ ์ฃผ์: repetition_penalty๋ฅผ 1.2 ์ด์์ผ๋ก ์ค์ ํ๋ฉด ํ๊ตญ์ด ํ ์คํธ ํ์ง์ด ์ ํ๋ ์ ์์ต๋๋ค. ์ด ๋ชจ๋ธ์ ๊ฒฝ์ฐ ๋๋ถ๋ถ์ ์ฌ์ฉ ์ฌ๋ก์์ 1.0-1.1์ด ์ต์ ์ ๋๋ค.
๊ณ ๊ธ ์ค์ ์์:
# ์ฝ๋ ์์ฑ์ฉ
code_gen_config = {
"max_new_tokens": 1024,
"temperature": 0.2,
"top_p": 0.9,
"repetition_penalty": 1.0,
"do_sample": True,
}
# ๋ํํ ์๋ต์ฉ
conversation_config = {
"max_new_tokens": 512,
"temperature": 0.7,
"top_p": 0.9,
"repetition_penalty": 1.05,
"do_sample": True,
}
# ์ ํํ ์ฌ์ค ๋ต๋ณ์ฉ
factual_config = {
"max_new_tokens": 256,
"temperature": 0.1,
"top_p": 0.85,
"repetition_penalty": 1.0,
"do_sample": True,
}
์ ํ์ฌํญ
์ด ๋ชจ๋ธ์ ์ฐ๊ตฌ ๋ฐ ๊ต์ก ๋ชฉ์ ์ผ๋ก ๊ณต๊ฐ๋์์ต๋๋ค. ์์ ์ ์ฌ์ฉ ์ CC-BY-NC-SA-4.0 ๋ผ์ด์ ์ค๋ฅผ ์ค์ํด์ผ ํฉ๋๋ค. ํ๊ตญ์ด์ ์ต์ ํ๋์ด ์์ผ๋ ๋ค๋ฅธ ์ธ์ด๋ ๋ถ๋ถ์ ์ผ๋ก ์ง์ํฉ๋๋ค.
๋ผ์ด์ ์ค
- ๋ชจ๋ธ ๋ผ์ด์ ์ค: CC-BY-NC-SA-4.0
- ๋ฒ ์ด์ค ๋ชจ๋ธ: EEVE-Korean-Instruct-10.8B-v1.0 ๋ผ์ด์ ์ค ์ค์
- ์์ ์ ์ฌ์ฉ: ์ ํ์ (๋ผ์ด์ ์ค ์ฐธ์กฐ)
์ธ์ฉ
@misc{eeve-vss-smh-2025,
author = {MyeongHo0621},
title = {EEVE-VSS-SMH: Korean Custom Fine-tuned Model},
year = {2025},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/MyeongHo0621/eeve-vss-smh}},
note = {LoRA fine-tuned and merged model based on EEVE-Korean-Instruct-10.8B-v1.0}
}
๊ฐ์ฌ์ ๊ธ
- ๋ฒ ์ด์ค ๋ชจ๋ธ: Yanolja - EEVE-Korean-Instruct-10.8B-v1.0
- ํ๋ จ ์ธํ๋ผ: KT Cloud H100E
- ํ๋ ์์ํฌ: Hugging Face Transformers, PEFT
์ฐ๋ฝ์ฒ
- Github : Fine-tuning-LLM
Last Updated: 2025-10-12
Checkpoint: 6250 steps (Epoch 1.94)
Training Method: Unsloth (2-5x faster)
Selection Criteria: Lowest Evaluation Loss (1.4604)
Status: Merged & Ready for Deployment
- Downloads last month
- 6
Model tree for MyeongHo0621/eeve-vss-smh
Base model
upstage/SOLAR-10.7B-v1.0