You can deploy this model and start earning money today!

paster0823/
m4ore-sport-slm-llama3.1-finetune

Name: m4ore-sport-slm-llama3.1-finetune
Availability: InStock

safetensors

llama

Views

No views yet

M4ore Sport Llama 3.1 Fine-tune

⚠️ 重要聲明 / Important Notice 本模型是基於 Llama-3.1-TAIDE-LX-8B-Chat 進行微調的衍生模型。 This model is a derivative work based on Llama-3.1-TAIDE-LX-8B-Chat.

Model Description

This model is a full parameter fine-tuned version of Llama-3.1-TAIDE-LX-8B-Chat specialized for sports and fitness-related question answering and text generation.

Base Model: taide/Llama-3.1-TAIDE-LX-8B-Chat
Model Size: 8B parameters
Fine-tuning Method: Full parameter fine-tuning
Training Hardware: Single H100 GPU
Training Duration: ~40 minutes
Training Data: 13M tokens of sports and fitness-related content
Model Type: Causal Language Model (LlamaForCausalLM)

Intended Use

This model is designed for:

Sports and fitness-related question answering
Text generation about sports and fitness topics
Conversational AI applications in the sports/fitness domain

Usage

python

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_name = "paster0823/m4ore-sport-slm-llama3.1-finetune"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

# Example prompt
prompt = "請問跑步前應該做什麼熱身運動？"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.7,
    top_p=0.9,
    do_sample=True
)

response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Model Details

Architecture: Llama 3.1 (32 layers, 4096 hidden size)
Attention Heads: 32
Key-Value Heads: 8 (GQA)
Max Position Embeddings: 131,072 tokens
Vocabulary Size: 188,256
Precision: bfloat16

Training Details

Base Model: TAIDE Llama 3.1 8B Chat
Training Tokens: 13 million
Hardware: NVIDIA H100 (single GPU)
Training Time: Approximately 40 minutes
Fine-tuning Type: Full parameter fine-tuning

Limitations

This model is specifically fine-tuned for sports and fitness domains. Performance on other topics may vary.
As with all language models, outputs should be verified for accuracy, especially for medical or health-related advice.
The model inherits limitations from the base Llama 3.1 model.

License

This model is released under the Llama-3.1-TAIDE Models License Agreement, which is based on:

Meta AI Llama 3.1 Community License Agreement
TAIDE-specific terms and conditions

Important License Requirements:

✅ Commercial use allowed (with certain restrictions)
✅ Modification and derivative works allowed
✅ Research and educational use allowed
❌ Military purposes prohibited
❌ Illegal purposes prohibited
⚠️ If you create derivatives, you must:
- Clearly indicate modifications made
- Provide this license to your users
- Prominently mark it as based on TAIDE models

Full License:

Governing Law: Republic of China (Taiwan) law applies to this license.

Citation

If you use this model, please cite:

bibtex

@misc{m4ore-sport-llama3.1,
  title={M4ore Sport Llama 3.1 Fine-tune},
  author={M4ore Team},
  year={2025},
  howpublished={\url{https://huggingface.co/paster0823/m4ore-sport-slm-llama3.1-finetune}}
}

Acknowledgements

This model is built upon:

Meta's Llama 3.1 architecture
TAIDE's Llama 3.1 Traditional Chinese adaptation