d10 final baseline (tok2B, eval@250, 18shards)

Completed
UUID:9d6fb60d-595c-47ca-913b-124c91d20eb5
Tao Linvia🐎 pentiumCreated 212 days agoUpdated 212 days ago

COMPLETED — d10 Final Baseline (tok2B, eval@250, 18shards)

MetricValue
val_bpb0.897
Training43.8 min
Wall60.7 min
MFU55.0%
Peak mem21.2 GB
Cost$4.36

Upstream defaults validated: tok_max_chars=2B, eval_every=250. 18 shards ensures single-epoch. Wall time fits in ~1hr budget.

v2 · Edited 212 days ago
Dryrun Job 
Funding 
Funded Job 
Done 

Finance

Fundings at snapshot$17.82

Individual funding and transaction records are not included in the public archive.

Configuration

Repository
https://github.com/mlpatron/mlpatron-nanochat
Version
experiment/baseline-search → 83770dbbcc82ebe554e6355eac3b8b5c76b017d3
Image
ghcr.io/nblintao/mlpatron-nanochat:latest
→ ghcr.io/nblintao/mlpatron-nanochat@sha256:28c5544c206237e407241b0a72a8707e63e52750f487b751675b51ff1bce1520
Entrypoint
a100_preset
Resources
NVIDIA A100 40GB × 1·a2-highgpu-1g (12 vCPU / 85 GiB)
Parameters
depth=10·eval_every=250·max_seq_len=2048·tok_max_chars=2000000000·num_iterations=1500(dryrun: 20)·num_data_shards=18·device_batch_size=32