alyzaki/smollm2-360m-dolly-lora

SmolLM2-360M Dolly-15k LoRA (Merged)

Fine-tuned SmolLM2-360M-Instruct on the full databricks/databricks-dolly-15k dataset using LoRA.

Model Repo

alyzaki/smollm2-360m-dolly-lora

Training Details

• Base Model: HuggingFaceTB/SmolLM2-360M-Instruct
• Dataset: Dolly-15k
• Context length: 256
• Epochs: 1
• Method: LoRA (merged)
• Hardware: Google Colab T4 GPU

Example Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("alyzaki/smollm2-360m-dolly-lora")
tokenizer = AutoTokenizer.from_pretrained("alyzaki/smollm2-360m-dolly-lora")