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

Expose eval_loss() and add a --eval-ckpt <path> branch to bin/train: load an
existing checkpoint into a model of the given arch and score it on the held-out
val split, then exit. Lets v0 and v1 be measured on the identical validation set
(the acceptance metric) without a separate eval binary.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-15 18:44:40 +08:00

..

src

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

2026-06-15 18:44:40 +08:00

tests

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

2026-06-15 18:34:48 +08:00

build.rs

data: gpt2 bpe via xserv-tokenizer + TinyStories corpus + lr schedule + grad clip

2026-06-15 16:29:32 +08:00

Cargo.toml

export: dump_logits bin for xserv-vs-xtrain comparison

2026-06-15 17:36:41 +08:00