atlas-sft / README.md
tocsa's picture
Upload README.md with huggingface_hub
748c184 verified
|
Raw History Blame Contribute Delete
2.4 kB
# tocsa/atlas-sft
**Atlas customer-support triage (capstone, Section 5)** — SFT model.
Atlas is a fictional SaaS project-management tool. The model reads an incoming support ticket and returns a `category`, an `urgency`, and a short drafted response — the Meridian pattern (eval -> SDG -> SFT -> DPO) applied end-to-end to a new domain during the capstone.
## Base model / architecture
nvidia/Mistral-NeMo-Minitron-8B-Instruct (MistralForCausalLM)
## Pipeline position
SFT model: a LoRA adapter trained on synthetically generated data, merged into the base model. This is the checkpoint used as the DPO starting point.
## Files
- `config.json` (702.0B)
- `generation_config.json` (111.0B)
- `model.safetensors` (15.7GB)
- `special_tokens_map.json` (414.0B)
- `tokenizer.json` (8.8MB)
- `tokenizer_config.json` (173.4KB)
## Recorded metrics
### baseline_eval.json
- **n**: 12
### best_sft.json
- **adapter_path**: /scratch/capstone_sft/checkpoints/epoch_2_step_56/model
- **epoch**: 2
## How to load
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("tocsa/atlas-sft")
tokenizer = AutoTokenizer.from_pretrained("tocsa/atlas-sft")
```
## Course context
Produced in the NVIDIA instructor-led workshop **"Adding New Knowledge to LLMs"** — Designing and building model-customization pipelines for domain workflows.
The workshop's build-time customization workflow:
| Stage | What it does | Tool |
|---|---|---|
| Task definition | taxonomy, schema, success criteria | |
| Eval | measure target behavior | vLLM |
| Gap evidence | baseline + ICL ceiling | |
| Synthetic data | targeted examples | NeMo Data Designer |
| Curation | quality + dedup | NeMo Curator |
| SFT | LoRA task behavior | NeMo AutoModel + PEFT |
| Serve + eval | compare model stages | vLLM |
| DPO | preference behavior | NeMo RL |
| Model artifact | adapter or DPO model | |
Open-model ecosystem used across the workshop:
| Models | Data | Training | Serving |
|---|---|---|---|
| Nemotron | NeMo Data Designer | NeMo AutoModel | vLLM |
| Open weights | NeMo Curator | LoRA / PEFT | NGC containers |
| Teacher models | Hugging Face | NeMo RL | Local eval |
Training data for this run was synthetically generated (NeMo Data Designer) and/or curated during the workshop exercises. Educational artifact — not reviewed for production use.