5 days ago
-
Card: BLiMP count note — cause of the 47 913 count not determined
-
Card: BLiMP count 47 912 (one pair differs at a numerical tie on another GPU)
-
Scenario chart: move decay-band label off the legend
-
Scenario chart: measured = solid/filled, projection = dashed/hollow
-
Scenario chart: continuation from checkpoint 1050 (before decay); remove resume-from-decayed scenario
-
Scenario chart: neutral label for the decay band (other LR budgets not yet decayed)
-
Add training-loss scenario chart (backtest failed; not a forecast); fix decay shape (1 - sqrt)
-
Card: Polish leaderboard columns (two independent implementations)
-
Card: final training facts and leaderboard-protocol results
-
Slayer1B 1.00 weights: step 1430 (ckpt sha 7d092d4d), exported 1:1 from the trainer checkpoint
-
Card: available vs planned tokens; no empty sections
-
Card: live training curve link
-
Model card: architecture, data, recipe; training in progress
-
initial commit
