smollm2-135m-noval-instruct/README.md

library_name, pipeline_tag, language, base_model, datasets, tags, license

library_name	pipeline_tag	language	base_model	datasets	tags	license
transformers	text-generation	en	atenareply/smollm2-135m-noval HuggingFaceTB/SmolLM2-135M	atenareply/noval-corp-sft-small HuggingFaceTB/smol-smoltalk	smollm2 sft instruction-tuning chatml	apache-2.0

SmolLM2-135M Noval — Instruct (SFT)

Instruction-tuned domain assistant: SFT of the 135M CPT model on grounded domain instruction pairs blended with general instructions.

Overview

• Stage: SFT (TRL SFTTrainer, ChatML, assistant-only loss)
• Lineage: SmolLM2-135M → CPT (noval) → SFT (this model)
• Method: SFT from the CPT'd checkpoint. ChatML with a {% generation %} block + assistant-only loss; grounded instruction pairs (teacher-distilled, generator≠judge, verbatim-number grounding filter [genie]) blended 60/40 with smol-smoltalk [collapse].
• Domain: fictional — Orbital Mining Corporation (OMC) technical docs + Mars Express telemetry.

Training

Dataset	noval-corp-sft-small — train 1,388 / val 43 (chat messages)
LR / epochs	2e-5 cosine, warmup 0.03, 3 epochs, max_len 2048, eff_batch 32

Evaluation

Metric	Value	Note
train_loss	2.257	healthy curve, no overfit
eval_loss	1.946	per-epoch 2.048 → 1.959 → 1.946
token accuracy	0.57	0.47 → 0.57

Eval is a held-out, deterministic verifiable harness (synthetic tasks); baseline = the pre-SFT ParamΔ model. See noval-corp/scripts/eval_agentic.py.

Intended use & limitations

Domain Q&A / summarization / code-explanation over OMC / Mars-Express, in chat (ChatML) format.

Limitations:

• 135M instruct is intrinsically limited.
• Instruction-following eval was handed off externally (no in-repo task metrics).
• Fictional domain; sparse one-off entities hallucination-prone.

Innovations tested

• Local-subscription teacher distillation — grounded pairs generated by Claude Code (generator ≠ judge), with a rule-based verbatim-number grounding gate [genie].

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("atenareply/smollm2-135m-noval-instruct")
model = AutoModelForCausalLM.from_pretrained("atenareply/smollm2-135m-noval-instruct")
msgs = [{"role": "user", "content": "What does the Orbital Mining Corporation do?"}]
inputs = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt", return_dict=True)
print(tok.decode(model.generate(**inputs, max_new_tokens=256)[0, inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Citations

• GENIE (grounded synthesis) — Mirza et al., ICLR 2024. arXiv:2401.14367
• Model collapse (≥50% real data) — Shumailov et al., 2024. arXiv:2404.01413

---

Card generated by noval-corp/scripts/gen_model_cards.py (standardized across the noval-corp model family).