Qwen2.5-7B-Inst-Math-Reasoning-SFT

Uploaded Model

• Developed by: Harsha901
• License: Apache-2.0
• Finetuned from model: unsloth/Qwen2.5-7B-Instruct

This Qwen2.5-7B model was fine-tuned using Unsloth for faster and more memory-efficient training, together with Hugging Face’s TRL library for supervised fine-tuning.

<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>

---

Model Overview

This is an instruction-tuned causal language model based on Qwen2.5-7B, designed to follow user prompts accurately and generate coherent, high-quality responses.

The model preserves the general-purpose strengths of Qwen2.5 while benefiting from domain-focused supervised fine-tuning.

---

Training Details

• Base model: Qwen2.5-7B-Instruct (Unsloth variant)
• Fine-tuning method: Supervised Fine-Tuning (SFT)
• Frameworks: Hugging Face Transformers + TRL
• Acceleration: Unsloth (2× faster training, reduced VRAM usage)
• Precision: FP16 / BF16 (hardware dependent)

---

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Harsha901/<YOUR-MODEL-NAME>"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    torch_dtype="auto"
)

---

Limitations

• Outputs may contain factual or reasoning errors
• Not intended for high-stakes or safety-critical applications
• Performance depends on prompt quality and context length

---

License

Released under the Apache 2.0 License, consistent with the base Qwen2.5 model.

---

Acknowledgements

• Qwen Team for the Qwen2.5 base model
• Unsloth for efficient fine-tuning optimizations
• Hugging Face for the training and hosting ecosystem