Views
No views yet
{"values":[...]}); solving = querying the policy cell-by-cell. Companion to the GPT-2 line in
Avra98/sudoku-gpt2-curriculum.--stage_i):| Stage | Target = values consistent under … | Latent steps k |
|---|---|---|
| 1 | direct row/column/box constraints | 1 |
| 2 | + 2-step lookahead | 2 |
| 3 | + 3-step lookahead | 3 |
--latent_mode recurrent_hidden). When training a
frontier stage, backtracking (--backtrack_enable) interleaves rehearsal of earlier stages
(decoded with their smaller latent budget), so the model does not forget earlier-stage skills:--remember_rate r — probability of broad rehearsal (sample any earlier stage uniformly);
otherwise sample from the earliest regressed stage up to the frontier.--backtrack_detect_threshold t — per-stage exact-match bar below which a stage counts as
regressed, driving targeted backtracking (t=0 ⇒ pure rehearsal).--backtrack_pool_rows N — cap the per-stage rehearsal pool size to keep dataset prep fast.s2_long/)| Run | remember_rate | detect_threshold | LR | Role |
|---|---|---|---|---|
ctrl_nobt / ctrl_nobt_lr2 | – | – | 5e-5 / 2e-5 | latent baseline (no backtracking) |
bt_rr03_adapt / _lr2 | 0.3 | 0.95 (adaptive) | 5e-5 / 2e-5 | targeted backtracking |
bt_rr05_adapt / _lr2 | 0.5 | 0.97 (adaptive) | 5e-5 / 2e-5 | broader targeted backtracking |
bt_rr05_warm / _lr2 | 0.5 | 0 (pure rehearsal) | 5e-5 / 2e-5 | pure rehearsal |
sft/, grpo_fixed/) and Stage-3 (s3_sft/, s3_grpo/) phases.
Backtracking is applied in the SFT phase; ctrl_nobt* are the no-backtracking controls.checkpoints/<run>/<phase>/<checkpoint>/ # LoRA adapters (+ tokenizer/config)
code/latent_multi_output_cell_policy/ # SFT + GRPO trainers
logs/ # training logs, pipeline + push scripts, RESULTS.mdQwen/Qwen2.5-1.5B-Instruct with PEFT.1# Stage-3 pipeline for a run (SFT k=3 -> GRPO k=3), optional GPU override:
2bash pipeline_stage3.sh <run_name> [gpu_id]