r1032-selective-fallback-v2-coverage (scale 1.00)
This is a standalone, merged BF16 checkpoint derived from
unconst/Affine-5czsc2fc98-r1032-vera-odpo-midrank-hibeta-shortctx-ultraextra-ep4-midlr-merged. It applies a
scaled selective-fallback LoRA trained on preserved positive turns and
sanitized, task-specific alternatives for high-confidence negative turns. It
does not require a runtime router, custom Python code, or a PEFT adapter.
Training summary
- Exact parent revision:
62dfb322fdce5873543bd92692ab4ecc3e13f941 - Adapter scale at merge:
1.00 - LoRA: r16, alpha64, dropout0.0, all-linear
- Objective: DPO, beta0.2, learning rate 1.0e-07, 1.0 epoch
- Context during training: 8192 tokens
- Training GPUs: 2 x NVIDIA H200
- Selected target mix before context filtering: 69% preserve / 31% fallback
Held-out preference proxy
The table compares the scaled adapter with the untouched R1032 parent. These small held-out metrics selected the merge strength; they are not a substitute for the exact full Affine duel on the dedicated evaluator.
| route | rows | mean reward margin | preference accuracy |
|---|---|---|---|
| fallback | 13 | 0.104621 | 0.6154 |
| preserve | 50 | 0.094390 | 0.5200 |
Qualification status
Experimental candidate. Before submission, run exact stock-vLLM Affine duels,
the exploit-pattern audit, repository preflight, and the official submission
client check. selective_fallback_provenance.json contains the machine-readable
training and merge record.