d12 A100 batch32 (OOM test)

Cancelled
UUID:a1e58f29-ba3e-4cfa-ab6c-9b63e736f0e6
Tao Linvia🐎 pentiumCreated 212 days agoUpdated 212 days ago

Test if d12 batch32 fits in A100 40GB. Estimated ~37GB peak mem — very tight. If OOM, batch16 is the fallback but much slower.

Dryrun Job 
—
Funding 
3
Funded Job 
4
Done 

Finance

Fundings at snapshot$0.00

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=12·eval_every=250·max_seq_len=2048·tok_max_chars=2000000000·num_iterations=2205(dryrun: 20)·num_data_shards=22·device_batch_size=32