amx-moe256-20b

Fine-tuned from /app/xlm/data/pretrain/m2r-init-e256-rb64-sig with ScalarXLM: version job_000029@step-305175, step 305175, full fine-tune.

Held-out loss at step 305175:

  • code_syn: 3.2288
  • facts: 3.1113
  • math: 3.1353
  • textbooks: 2.7780
  • web: 4.5132
  • wiki: 3.9083
Downloads last month
20
Safetensors
Model size
0.5B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for gdiamos/amx-moe256-20b

Finetunes
2 models

Collection including gdiamos/amx-moe256-20b