Rouslan Zenetl rzen
  • Joined on 2019-07-14
rzen pushed to main at rzen/blueprints 2026-09-26 02:52:54 +00:00
bca2e3f5a0 Silence the stove detector's queue overflow and fix notify calls in stove and HVAC blueprints
rzen pushed to main at rzen/blueprints 2026-09-06 11:11:29 +00:00
591e66b208 Use the abs filter instead of an abs() call in the stove and HVAC blueprints
rzen pushed to main at rzen/experiment-llm 2026-07-12 22:28:45 +00:00
921069c5db Add degradation analysis + backing artifacts
rzen pushed to main at rzen/experiment-llm 2026-07-12 16:58:50 +00:00
9fd56b6063 Add degradation experiments: one-parameter-at-a-time ablations
rzen pushed to main at rzen/experiment-llm 2026-07-12 16:44:13 +00:00
84c41bc612 Phase 5: v1 sample gallery — iteration 1 complete
rzen pushed to main at rzen/experiment-llm 2026-07-12 16:36:04 +00:00
116e1f61a3 Mark full training run item done (val loss 1.52)
rzen pushed to main at rzen/experiment-llm 2026-07-12 16:35:49 +00:00
ba73c8a7e8 Phase 4: v1 training run results (val loss 1.52, ppl ~4.6)
rzen pushed to main at rzen/experiment-llm 2026-07-12 15:34:56 +00:00
72595e274b Mark overfit sanity check + documented-commands items done
rzen pushed to main at rzen/experiment-llm 2026-07-12 15:34:47 +00:00
8915d274f5 Phase 5: autoregressive sampler (temperature + top-k)
rzen pushed to main at rzen/experiment-llm 2026-07-12 15:32:35 +00:00
4947556c02 Phase 4: dependency-free ASCII loss-curve plotter
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:51:59 +00:00
234a37f8fd Phase 4: training loop (code only — for user to run)
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:49:09 +00:00
7b25fc89f6 Mark Phase 2 encoding + Phase 3 model items done
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:48:57 +00:00
e708a5b886 Phase 3: decoder-only transformer (~4.3M params)
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:38:23 +00:00
c202bada6a Phase 2: token dataset + corpus encoder (code only, not yet run)
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:33:35 +00:00
6111aac459 Mark Phase 1 tokenizer item done
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:33:26 +00:00
3a489fef0a Phase 1: trained 4096-vocab tokenizer + per-word encode cache
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:31:19 +00:00
2c5fc29592 Phase 1: from-scratch byte-level BPE tokenizer
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:26:05 +00:00
e760c00043 Mark Phase 0 environment + dataset item done
rzen pushed to main at rzen/experiment-llm 2026-07-12 14:25:54 +00:00
0c26cd1f1f Record canonical TinyStoriesV2 SHA-256 hashes for reproducibility
rzen created branch main in rzen/experiment-llm 2026-07-12 14:24:54 +00:00