Deploy audited RTX 6000D Flash-Next 128K multimodal baseline
This commit is contained in:
@@ -0,0 +1,14 @@
|
||||
[
|
||||
{
|
||||
"file": "v1/worker/gpu/model_states/mamba_hybrid.py",
|
||||
"sha256": "573e130ac5587d4822fabf97adad1ad0fda02ff83c5e70df879a5bb175ae845c",
|
||||
"before": "(new_req_data.num_computed_tokens - 1) // self.cache_config.block_size",
|
||||
"after": "(new_req_data.num_computed_tokens - 1)\n // (self.cache_config.mamba_block_size or self.cache_config.block_size)"
|
||||
},
|
||||
{
|
||||
"file": "v1/core/sched/scheduler.py",
|
||||
"sha256": "ee5800550e8d52ddc81c9770bb82cf0c9a47a22a96d44f6e6f7d2e67d9900795",
|
||||
"before": " block_size = self.cache_config.block_size\n # The last block-aligned",
|
||||
"after": " block_size = self.block_size\n # The last block-aligned"
|
||||
}
|
||||
]
|
||||
Reference in New Issue
Block a user