diff options
| author | Paul Buetow <paul@buetow.org> | 2026-03-18 18:06:42 +0200 |
|---|---|---|
| committer | Paul Buetow <paul@buetow.org> | 2026-03-18 18:06:42 +0200 |
| commit | 07f91d85eb7d115ccfbecb9841712a12d36e874e (patch) | |
| tree | 3f9c2db006ae22dd7a5deb8d243675fdb32b09c7 /snippets/hyperstack/hyperstack.rb | |
| parent | 1122c9373cadb90d28b8d588e73f84b86237fd15 (diff) | |
nemotron-super: set max_model_len=262144 (256K); document NoPE and OOM risk
Tested 1M context (NoPE allows arbitrary max_position_embeddings without
YaRN) — OOMs on A100 80GB due to insufficient VRAM after 60GB model weights.
256K (262144) is the practical ceiling on this hardware.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Diffstat (limited to 'snippets/hyperstack/hyperstack.rb')
0 files changed, 0 insertions, 0 deletions