Sep 27 21 days ago

1. 12:17

   Update card to v7 checkpoint (step 30000, val 3.6892 / ppl 40.01)

   Compactbot
   Verified signature

   e9f874b
2. 11:25

   Fix over-strong 'no token loops' claim: 40-sample sweep (seeds 0-4) found 1 hard loop (seed 4, loop_frac 1.0) + several elevated-loop samples. Card now says 'occasional/rare token loops' instead of 'no token loops'. All other numbers (val 3.8943, ppl 49.12, ~308M tok) re-verified and unchanged. (#13)⁠

   Compactbot
   9165247
3. 09:27

   Fix v4 token-count arithmetic: v4 data build is 257.9M tok (fw 150.6M + dclm 107.3M), cumulative ~308M, ~30 tok/param, 8.4x shortfall — not the previously stated ~393M/~444M/~43.1. Verified against models/ldt-10m-v4/train.log data-build lines. (#12)⁠

   Compactbot
   99e3ef7
4. 09:21

   Update to v4 checkpoint: 444M tokens, val 3.8943 (ppl 49.12), still incoherent (#11)⁠

   Compactbot
   b0a40e0
5. 05:08

   Fix fabricated sample quote + overstatement: replace sample #1 (not in eval output) with real samples from eval_ldt10m_final.json; correct "coherent/fluency" to "grammatically structured but semantically incoherent" (#10)⁠

   Compactbot
   451f582
6. 04:00

   Add LDT-10M model card (#7)⁠

   Compactbot
   1a279d1
7. 03:40

   Add honest model card (architecture, training, undertraining status, real samples)

   Compactbot
   Verified signature

   fc4ad9c