metadata
license: apache-2.0
tags:
- llama
- unsloth
- gguf
- quantized
- lora
- instruction-tuning
- colab
datasets:
- custom
language:
- en
library_name: unsloth
pipeline_tag: text-generation
Playwright1 GGUF Model
This model is a 4-bit LoRA fine-tuned version of unsloth/Llama-3.2-3B-Instruct, optimized for conversational instruction-following tasks. Trained on custom command-response data using the ShareGPT format.
Features
- 🧠 Fine-tuned with LoRA (r=16) using Unsloth
- 💾 Quantized to 4-bit (q4_k_m) for fast inference
- 🔧 Ideal for lightweight deployment
Training Info
- Trained with
SFTTrainer(TRL) for 60 steps with 2 batch size on Google Colab.
License
- Apache 2.0