ThakrePranjal/pharma-instruction-dataset-unsloth
Pharma Instruction Dataset — Unsloth Pipeline Instruction-tuning dataset in Alpaca format used for Stage 2 SFT in the Unsloth 3-stage pharma fine-tuning pipeline. Format { "instruction": "Explain the primary mechanism of action of metformin.", "input": "", "output": "Metformin primarily acts by activating AMP-activated protein kinase (AMPK)..." } Training prompt template ### Instruction: <instruction> ### Input: (omitted if empty)… See the full description on the dataset page: https://huggingface.co/datasets/ThakrePranjal/pharma-instruction-dataset-unsloth.
09
Pharma Instruction Dataset — Unsloth Pipeline
Instruction-tuning dataset in Alpaca format used for Stage 2 SFT in the Unsloth 3-stage pharma fine-tuning pipeline.
Format
{
"instruction": "Explain the primary mechanism of action of metformin.",
"input": "",
"output": "Metformin primarily acts by activating AMP-activated protein kinase (AMPK)..."
}Training prompt template
### Instruction:
<instruction>
### Input: (omitted if empty)
<input>
### Response:
<output>Stats
- Total rows: 48
- Train/eval: used as-is for SFT (no split; packing=False)
Pipeline context
Stage 1 (SFT - Domain Pretraining) → raw PDF paragraphs
Stage 2 (SFT - Instruction Tuning) → THIS DATASET
Stage 3 (DPO - Preference Tuning) → [ThakrePranjal/pharma-preference-dataset-unsloth](https://huggingface.co/datasets/ThakrePranjal/pharma-preference-dataset-unsloth)Related models
- Stage 2 LoRA adapter : ThakrePranjal/pharma-tinyllama-unsloth-stage2-lora
- Stage 2 merged model : ThakrePranjal/pharma-tinyllama-unsloth-stage2-merged
