Skip to content

misc: correct the int data type for token ids and indices #969

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zhyncs merged 1 commit into from
Aug 7, 2024
Merged

misc: correct the int data type for token ids and indices #969

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion python/sglang/srt/managers/schedule_batch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -780,7 +780,7 @@ def top_k_top_p_sampling_from_probs_torch(
sampled_index = torch.multinomial(probs_sort, num_samples=1)
except RuntimeError:
batch_next_token_ids = torch.zeros(
(probs_sort.shape[0],), dtype=torch.int64, device=probs.device
(probs_sort.shape[0],), dtype=torch.int32, device=probs.device
)
success = torch.zeros(probs.shape[0], dtype=torch.bool, device=probs.device)
return batch_next_token_ids, success
Expand Down
4 changes: 2 additions & 2 deletions python/sglang/srt/mem_cache/radix_cache.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -74,7 +74,7 @@ def match_prefix(self, key, **kwargs):
if value:
value = torch.concat(value)
else:
value = torch.tensor([], dtype=torch.int64)
value = torch.tensor([], dtype=torch.int32)
return value, last_node[0]

def insert(self, key, value=None):
Expand Down Expand Up @@ -102,7 +102,7 @@ def cache_req(
if del_in_memory_pool:
self.token_to_kv_pool.free(indices)
else:
return torch.tensor([], dtype=torch.int64), self.root_node
return torch.tensor([], dtype=torch.int32), self.root_node

# Radix Cache takes one ref in memory pool
self.token_to_kv_pool.free(indices[last_uncached_pos:new_prefix_len])
Expand Down