tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Fixes comments from winglian

Browse Source
This commit is contained in:
mhenrhcsen
2025-05-12 22:43:15 +02:00
parent be3c6bbd85
commit 5dd8f0b2b8
7 changed files with 262 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
docs/config.qmd
View File

@@ -332,8 +332,8 @@ dataset_shard_idx:
# The maximum length of an input to train with, this should typically be less than 2048
# as most models have a token/context limit of 2048
sequence_len: 2048
# How to handle tokens exceeding max sequence length - "drop" (default, removes sample) or "truncate" (cuts off excess tokens)
excess_token_handling: drop
# How to handle sequences that overflow the sequence_len: 'drop' (default, removes sample) or 'truncate' (cuts off excess tokens).
sequence_len_overflow_handling: drop
# Pad inputs so each step uses constant sized buffers
# This will reduce memory fragmentation and may prevent OOMs, by re-using memory more efficiently
pad_to_sequence_len: