Downloads · 30 days
11
14% of all-time downloads
burtenshaw/lfm25-350m-commonsenseqa-sft-lr1e4-r16-2k
lfm25-350m-commonsenseqa-sft-lr1e4-r16-2k is a text generation model from burtenshaw. Use it when you need the model to write or continue text. It is set up for peft.
This is a PEFT LoRA adapter for LiquidAI/LFM2.5-350M, trained with TRL SFT on tau/commonsenseqa train examples.
Downloads · 30 days
11
14% of all-time downloads
All-time downloads
76
Public
Repo size
72.2 MB
Likes
0
Public
Click a slice to open those files.
.safetensors24 MB · 81%
From the Hugging Face model README
This is a PEFT LoRA adapter for LiquidAI/LFM2.5-350M, trained with TRL SFT on tau/commonsense_qa train examples.
The prompt format is a question, labeled answer choices, and Answer:. The completion target is the single correct letter. Evaluation scores each candidate letter by summed log probability and chooses the highest-scoring answer.
Same HF Job computed base and adapter accuracy on the full CommonsenseQA validation split (1221 examples).
| Model | Forced-choice validation accuracy | Correct / Total | Job |
|---|---|---|---|
LiquidAI/LFM2.5-350M | 0.20393120393120392 | 249 / 1221 | 6a358b823093dba73ce2a757 |
LiquidAI/LFM2.5-350M + this adapter | 0.25225225225225223 | 308 / 1221 | 6a358b823093dba73ce2a757 |
Absolute gain: +0.048321048321048304.
A previous successful-but-unpushed run scored 0.2571662571662572 after SFT, but failed before artifact upload when post-train Trackio logging raised. The released checkpoint above is from the completed upload job.
SFTTrainer with completion-only loss and PEFT LoRA.LiquidAI/LFM2.5-350M at commit 7728373d9f752dc3669ee3bf70786aef397874bb.tau/commonsense_qa, train split capped at 2000 examples.tau/commonsense_qa validation split, all 1221 examples.1e-4, best checkpoint by validation eval_loss.6a358b823093dba73ce2a757.burtenshaw/lfm25-commonsenseqa-sft-trackio.Saved benchmark files in this repo:
benchmark_before.jsonbenchmark_after.jsonbenchmark_summary.jsonThis adapter is a focused CommonsenseQA SFT experiment, not a general reasoning model. Evaluate separately before using it for other tasks.