A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 11:15 UTC
Make JAIS compatible with Transformers v5 (#34264)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Harry Mellor committed
Feb 11, 2026 at 13:30 UTC
0f5e55e7a8de564407ee54ad8ab5ab1d2cb3bb5a
1 file changed
-1
vllm/model_executor/models/jais.py
-1
@@ -244,7 +244,6 @@ class JAISModel(nn.Module):
244
quant_config = vllm_config.quant_config
245
246
self.config = config
247
- assert not config.add_cross_attention
247
assert not config.scale_attn_by_inverse_layer_idx
248
assert not config.reorder_and_upcast_attn
249
self.embed_dim = config.hidden_size