-
921069c5db
Add degradation analysis + backing artifacts
main
rzen
2026-07-12 18:28:43 -04:00
-
9fd56b6063
Add degradation experiments: one-parameter-at-a-time ablations
rzen
2026-07-12 12:58:48 -04:00
-
84c41bc612
Phase 5: v1 sample gallery — iteration 1 complete
rzen
2026-07-12 12:44:10 -04:00
-
116e1f61a3
Mark full training run item done (val loss 1.52)
rzen
2026-07-12 12:36:03 -04:00
-
ba73c8a7e8
Phase 4: v1 training run results (val loss 1.52, ppl ~4.6)
rzen
2026-07-12 12:35:47 -04:00
-
72595e274b
Mark overfit sanity check + documented-commands items done
rzen
2026-07-12 11:34:55 -04:00
-
8915d274f5
Phase 5: autoregressive sampler (temperature + top-k)
rzen
2026-07-12 11:34:45 -04:00
-
4947556c02
Phase 4: dependency-free ASCII loss-curve plotter
rzen
2026-07-12 11:32:34 -04:00
-
234a37f8fd
Phase 4: training loop (code only — for user to run)
rzen
2026-07-12 10:51:58 -04:00
-
7b25fc89f6
Mark Phase 2 encoding + Phase 3 model items done
rzen
2026-07-12 10:49:06 -04:00
-
e708a5b886
Phase 3: decoder-only transformer (~4.3M params)
rzen
2026-07-12 10:48:55 -04:00
-
c202bada6a
Phase 2: token dataset + corpus encoder (code only, not yet run)
rzen
2026-07-12 10:38:20 -04:00
-
6111aac459
Mark Phase 1 tokenizer item done
rzen
2026-07-12 10:33:33 -04:00
-
3a489fef0a
Phase 1: trained 4096-vocab tokenizer + per-word encode cache
rzen
2026-07-12 10:33:24 -04:00
-
2c5fc29592
Phase 1: from-scratch byte-level BPE tokenizer
rzen
2026-07-12 10:31:16 -04:00
-
e760c00043
Mark Phase 0 environment + dataset item done
rzen
2026-07-12 10:26:03 -04:00
-
0c26cd1f1f
Record canonical TinyStoriesV2 SHA-256 hashes for reproducibility
rzen
2026-07-12 10:25:53 -04:00
-
5a7f2177bc
Phase 0: project plan, pinned environment, data download + MPS smoke test
rzen
2026-07-12 10:24:52 -04:00