Fix UniversalTransformerCache.get_mask_sizes for batched generation
1
#5 opened 1 day ago
by
KristianS7
Fix bos/eos token IDs + add enable_thinking to chat template
1
#4 opened 1 day ago
by
KristianS7
Update rope embeddings for rope_type='default'
#3 opened 2 days ago
by
sirorezka
Added pad token for the model
#2 opened 2 days ago
by
sirorezka