huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-12 12:54:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: update sliding window conditional

Browse Source
This commit is contained in:
drbh 2024-08-02 17:39:08 +00:00
parent 5b649d67c4
commit cf27954257
1 changed files with 3 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
server/text_generation_server/models/__init__.py
View File

@ -484,14 +484,13 @@ def get_model(
) )
sliding_window = config_dict.get("sliding_window", -1) sliding_window = config_dict.get("sliding_window", -1)
is_max_input_within_sliding_window = ( if max_input_tokens <= sliding_window:
max_input_tokens <= sliding_window if max_input_tokens is not None else False sliding_window = -1
)
if ( if (
(sliding_window is not None and sliding_window != -1) (sliding_window is not None and sliding_window != -1)
and not SUPPORTS_WINDOWING and not SUPPORTS_WINDOWING
and is_max_input_within_sliding_window and max_input_tokens > sliding_window
): ):
raise ValueError( raise ValueError(
f"The backend {SYSTEM} does not support sliding window attention that is used by the model type {model_type}. To use this model nonetheless with the {SYSTEM} backend, please launch TGI with the argument `--max-input-tokens` smaller than sliding_window={sliding_window} (got here max_input_tokens={max_input_tokens})." f"The backend {SYSTEM} does not support sliding window attention that is used by the model type {model_type}. To use this model nonetheless with the {SYSTEM} backend, please launch TGI with the argument `--max-input-tokens` smaller than sliding_window={sliding_window} (got here max_input_tokens={max_input_tokens})."