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

Migrate QAT API; fix axolotl quantize for QAT-ed models; add NVFP4 (#3107)

Browse Source
This commit is contained in:
salman
2025-09-12 10:55:50 +01:00
committed by GitHub
parent 0401a15888
commit 58d67bf98d
16 changed files with 554 additions and 339 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/e2e/test_qat.py
View File

@@ -43,7 +43,7 @@ class TestQATLlama:
"qat": {
"quantize_embedding": True,
"activation_dtype": "int8",
"weight_dtype": "int8",
"weight_dtype": "int4",
"group_size": 8,
},
"num_epochs": 1,
@@ -111,7 +111,7 @@ class TestQATLlama:
"qat": {
"quantize_embedding": True,
"activation_dtype": "int8",
"weight_dtype": "int8",
"weight_dtype": "int4",
"group_size": 8,
},
"save_first_step": False,