tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

fix: gemma3_text model loading vision config (#3354)

Browse Source 
* fix: gemma3-text mode loading vision config

* fix: improve defaults to use lora kernels
This commit is contained in:
NanoCode012
2026-01-13 21:49:23 +07:00
committed by GitHub
parent 258ce8d4fa
commit 359b7ad85e
6 changed files with 24 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
examples/gemma3/gemma-3-1b-qlora.yml
View File

@@ -1,6 +1,7 @@
base_model: google/gemma-3-1b-it
model_type: Gemma3ForCausalLM
cls_model_config: Gemma3TextConfig
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name
@@ -29,7 +30,7 @@ output_dir: ./outputs/out
adapter: qlora
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_dropout: 0
lora_target_linear: true
sequence_len: 2048