Fine-tuned Llama 3.2 3B to match 70B models for local transcripts

Show HN: Fine-tuned Llama 3.2 3B to match 70B models for local transcripts

Based, Fine-tuned Llama 3.2 3B to match 70B models for local transcripts works in show hn launch.

Profile

I wrote a small local tool to transcribe audio notes (Whisper/Parakeet). Code: https://github.com/bilawalriaz/lazy-notesI wanted to process raw transcripts locally without OpenRouter. Llama 3.2 3B with a prompt was decent but incomplete, so I tried SFT. I fine-tuned Llama 3.2 3B to clean/analyze dictation and emit structured JSON (title, tags, entities, dates, actions).Data: 13 real memos → Kimi K2 gold JSON → ~40k synthetic + gold; keys canonicalized. Chutes.ai (5k req/day).Training: RTX 4090 24GB, ~4h, LoRA (r=128, α=128, dropout=0.05), max seq 2048, bs=16, lr=5e-5, cosine, Unsloth. On 2070 Super 8GB it was ~8h.Inference: merged to GGUF, Q4_K_M (llama.cpp), runs in LM Studio.Evals (100-sam

As a business in show hn launch, Fine-tuned Llama 3.2 3B to match 70B models for local transcripts is listed here with the public details available about it.

It is recorded at the Show HN launch stage.

Website: bilawal.net/post/finetuning-llama32-3b-for-transcripts