vlm: increase the default max_num_batched_tokens for multimodal models
#1458
| Job | Run time |
|---|---|
| 16s | |
| 6s | |
| 8s | |
| 8s | |
| 9s | |
| 47s |
max_num_batched_tokens for multimodal models
#1458
| Job | Run time |
|---|---|
| 16s | |
| 6s | |
| 8s | |
| 8s | |
| 9s | |
| 47s |