GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
abhishekai
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
abhishekai
Model Tree
abhishekai/slm-500m-legal-sft
Input Modalities
Output Modalities
Supported Functionality
This model performs at parity with the SFT model it was trained from.
| win-rate vs slm-500m-legal-sft | 0.480 |
| decisive comparisons | 29 wins / 33 losses / 38 ties (n=100) |
| two-sided binomial p | 0.704 |
| 95% CI on decisive win-rate | [0.344, 0.592] |
The confidence interval straddles 0.5. DPO here did not measurably improve the model, and did not measurably harm it either. Published as a reproducible RLAIF artifact, not as an upgrade. If you want the better-understood model, use the SFT one.
Scored by Claude Sonnet on four axes out of 10 — a different model family from the Gemini judge behind the win-rate above, and one that had no hand in writing this project's training data. 300 held-out prompts, the same prompts and the same scale used for all twelve checkpoints, so this number is comparable across models in a way no win-rate here is.
| mean score | 5.08 / 10 |
| 95% CI (bootstrap over per-prompt scores) | [4.75, 5.41] |
| paired step | +0.42 against its own SFT (4.65), CI [+0.10, +0.75] - a real improvement, and the best score any model in this project achieves below Gemma scale. |
The full twelve-checkpoint table, with every stage-to-stage interval, is in
MODEL_INDEX.md in the project repository.
Because it is part of a controlled 3-model scaling comparison, run with an identical pipeline (same prompt pool, same Gemini 2.5 Flash judge, same preference-mining thresholds, same DPO recipe):
| model | params | DPO win-rate vs its own SFT |
|---|---|---|
| 125M | 126M | 0.455 (degraded) |
| 500M (this) | 518M | 0.480 (parity) |
| Gemma-2-2B | 2,614M | 0.530 (parity, trending positive) |
DPO's win-rate rises monotonically with model scale, crossing from harmful to parity to slightly positive. No individual point is statistically distinguishable from 0.5 at n=100 — the trend is the finding, not any single number. Distinguishing a 0.53 effect from parity would need roughly 400–600 eval prompts.
994 triplets (894 train / 100 held-out eval), built on-policy:
chosen really beats
rejected, with A/B order randomized to cancel position biasFunnel: 1,498 prompts → 1,380 with ≥2 distinct candidates → 1,048 cleared the score gap (76%) → 1,011 confirmed by the independent judge (96.5%) → 994 after dedup. That 96.5% agreement rate was the highest of the three models.
DPO (Rafailov et al. 2023) with an RPO-style NLL anchor on the chosen response. beta 0.1, LR 5e-6, 2 epochs, effective batch 16, 110 steps, 1×H100, 12.1 min, $0.80. Final eval preference accuracy 0.717, eval margin 0.689.
Identical to the SFT model — grounded prompts only, greedy or low temperature. See slm-500m-legal-sft.
All of the SFT model's limitations apply (grounded-only, no arithmetic reliability, synthetic training data, 1,024-token context, not legal or financial advice), plus:
Apache-2.0, inherited from thesreedath/slm-500m-base via slm-500m-legal-sft.