6 days ago
-
Card: drop v7
-
Remove superseded v7 (PyTorch Native Beta 5) cache
7 days ago
-
Card: v9 trn2 LNC=2 cache
-
v9 trn2 cache: SDK 2.33, LNC=2 with atom graph break (new trn2 default)
8 days ago
-
Card: add trn2 v8 cache
-
v8 trn2 cache: Neuron SDK 2.33, trn2.3xlarge LNC=1
-
Remove trn2 LNC=2 bf16 cache (accuracy issue found)
-
v8 trn2 cache: Neuron SDK 2.33, trn2.3xlarge LNC=2
9 days ago
-
Card: v8 cache for Neuron SDK 2.33
10 days ago
-
v8 cache: Neuron SDK 2.33 public DLAMI (torch-neuronx 2.14.3)
15 days ago
-
Dataset card: update stack wording
21 days ago
-
Card: v7 is the only file
-
Remove superseded caches v1-v6; v7 is the only supported file
22 days ago
-
Card: current file only
-
v7: caches for the trimmed public release
23 days ago
-
Card: point at v6, list files, exact shape coverage
-
v6: NEFF + AOT + precompile package built on inf2.xlarge (ds 1/8/25)
-
v5: NEFF + AOTAutograd + dynamo precompile package (ds 1/8/25)
-
v4: NEFF + bundled AOTAutograd cache (skips per-process re-lowering)
24 days ago
-
v3 cache: built from a setup_inf2.sh-only install (v2 key mismatch fix)
-
v2 cache: nonzero count shim + disto drop (ds 1/8/25, N=192)
27 days ago
-
Upload boltz_neuron_cache_ds1-8-25.tgz with huggingface_hub
-
Upload README.md with huggingface_hub
-
initial commit
