You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have read the README and searched the existing issues.
System Info
llama factory version: 0.9.1.dev0
platform : linux ubuntu
python: 3.11
Reproduction
nothing
Expected behavior
No response
Others
My understanding is that the cutoff_len parameter refers to the truncation of the input, without considering the model's output. This is based on the parameter description I saw in the code:
"The cutoff length of the tokenized inputs in the dataset."
However, the following discussion here seems to imply that the cutoff_len parameter is the sum of max_source_len and max_target_len. This is making me a bit confused.
This discussion was converted from issue #5820 on October 25, 2024 08:32.
Heading
Bold
Italic
Quote
Code
Link
Numbered list
Unordered list
Task list
Attach files
Mention
Reference
Menu
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Reminder
System Info
llama factory version: 0.9.1.dev0
platform : linux ubuntu
python: 3.11
Reproduction
nothing
Expected behavior
No response
Others
My understanding is that the cutoff_len parameter refers to the truncation of the input, without considering the model's output. This is based on the parameter description I saw in the code:
"The cutoff length of the tokenized inputs in the dataset."
However, the following discussion here seems to imply that the
cutoff_len
parameter is the sum ofmax_source_len
andmax_target_len
. This is making me a bit confused.Can someone help explain this?
Beta Was this translation helpful? Give feedback.
All reactions