blob: 6ff07bab72f7fb5be8be61976e89473628ba5a7a (
plain)
1
2
3
4
5
6
7
8
9
10
11
|
{
"author": "Renier Velazco",
"github_id": "chonchiog",
"name": "FP16 Tied Embedding + LR/Warmdown Tuning",
"blurb": "Keep tok_emb.weight in fp16 during int8 quantization to eliminate the output-head quantization gap (0.007 -> 0.0005 BPB). Slightly reduce MLP hidden (992 vs 1024) to fit within 16MB. Tune warmdown (3600 vs 1200) and matrix LR (0.06 vs 0.04) for better convergence under the 10-min wallclock cap.",
"date": "2026-03-18",
"val_loss": 2.05945460,
"val_bpb": 1.21972502,
"bytes_total": 15896222,
"bytes_code": 48125
}
|
