Gahow Wang

gahow deleted branch t10-validate from gahow/xtrain

2026-06-15 16:45:58 +00:00

gahow pushed to main at gahow/xtrain

2026-06-15 16:44:58 +00:00

9a25616a30 docs: Phase T10 — batched forward

4ccab0fb42 perf: KI-1 fixed — GPU util 0-15%→37-54%, tok/s 1653→25627 (15.5x)

25b032445d train: real batched step (drop loop+SUM)

5353b38402 model: batched forward [B,S]

7821bd9c34 autograd: batch dim for ops (flatten linears, batched attention)

Compare 5 commits »

gahow pushed to t10-validate at gahow/xtrain

2026-06-15 16:31:25 +00:00

67687ec8fe wip: T10 batched forward (validation)

gahow created branch t10-validate in gahow/xtrain

2026-06-15 16:19:32 +00:00

gahow pushed to t10-validate at gahow/xtrain

2026-06-15 16:19:32 +00:00

ce9d22ffc2 wip: T10 batched forward (validation)

gahow pushed to main at gahow/xtrain

2026-06-15 15:20:35 +00:00

d2a585c5cb docs: KI-1 re-diagnosed in v3 — larger batch does NOT fix DDP weak scaling

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 15:01:24 +00:00

b1b74318f6 Pin 27B A/B to GPUs 2-7 (route around leaked GPU0/1 memory)

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 14:58:30 +00:00

2fcaf80450 Wrap socket/timeout errors in HTTP client as HttpClientError

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 14:40:45 +00:00

3541065675 Speed up 27B TP A/B: request_timeout 180s, search.high 0.125

gahow pushed to main at gahow/xtrain

2026-06-15 14:38:38 +00:00

bf679f6f1f docs: run v2 — TinyStories, dim384/12L, DDP 4-card (val 1.71)

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 12:35:24 +00:00

7678c7d5e8 Switch 27B TP A/B to length-aware TTFT SLO (4s + L_in/8k), widen search

ed2bbe0323 Add linear_ms SLO rule (length-aware TTFT budget)

Compare 2 commits »

gahow pushed to main at gahow/xtrain

2026-06-15 11:57:00 +00:00

c87a0bc44e docs: known-issues / perf backlog — KI-1 DDP weak scaling at small global batch

gahow pushed to main at gahow/xtrain

2026-06-15 11:34:47 +00:00

7090b475fb train: bring DDP trainer to parity with bin/train (val + checkpoint + cache + arch)

gahow pushed to main at gahow/xtrain

2026-06-15 11:09:53 +00:00

264660527f docs: run v1 — TinyStories full, dim256

gahow pushed to main at gahow/xtrain

2026-06-15 10:46:25 +00:00

ec8114ecbc train: --eval-ckpt eval-only mode (v0-vs-v1 same-set val loss)

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 10:40:56 +00:00

77af4ded2a Flag Stop-B e2e per-GPU trajectory as non-benchmark (saturation + smoke regime)

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 10:39:55 +00:00

4f45b546a1 Add 27B TP A/B (deterministic ground-truth: does TP2 beat TP1 per-GPU)

gahow pushed to main at gahow/xtrain

2026-06-15 10:34:54 +00:00

e44e50ef78 data: full TinyStories + tokenized-id cache, val loss, CLI arch

15f1e526c7 train: parameterize model size (scaling ladder)

Compare 2 commits »

gahow pushed to feat/two-stop at gahow/aituner

2026-06-15 09:58:46 +00:00

90c3eb51c8 Document Stop-B end-to-end validation (Phase 5)

gahow pushed to main at gahow/xtrain

2026-06-15 09:37:48 +00:00

8981cf7982 docs: T9 verification results (xserv == xtrain, dash5)