ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Qwen2

Browse source
This commit is contained in:
Daniel Han-Chen 2024-06-07 02:53:17 +10:00
commit 23e4103895
2 changed files with 36 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

10
unsloth/models/llama.py
View file

@ -1782,10 +1782,12 @@ class FastLlamaModel:
layer.self_attn.apply_qkv = apply_lora_qkv
n_qkv += 1
else:
logger.warning_once(
"Not an error, but Unsloth cannot patch Attention layers with our manual autograd engine since either LoRA adapters\n"\
"are not enabled or a bias term (like in Qwen) is used."
)
if model_type != "qwen2":
logger.warning_once(
"Not an error, but Unsloth cannot patch Attention layers with our manual autograd engine since either LoRA adapters\n"\
"are not enabled or a bias term (like in Qwen) is used."
)
pass
pass
# O attention patching

30
unsloth/models/mapper.py
View file

@ -156,6 +156,36 @@ __INT_TO_FLOAT_MAPPER = \
"unsloth/Phi-3-medium-4k-instruct",
"microsoft/Phi-3-medium-4k-instruct",
),
"unsloth/Qwen2-0.5B-bnb-4bit" : (
"unsloth/Qwen2-0.5B",
"Qwen/Qwen2-0.5B",
),
"unsloth/Qwen2-0.5B-Instruct-bnb-4bit" : (
"unsloth/Qwen2-0.5B-Instruct",
"Qwen/Qwen2-0.5B-Instruct",
),
"unsloth/Qwen2-1.5B-bnb-4bit" : (
"unsloth/Qwen2-1.5B",
"Qwen/Qwen2-1.5B",
),
"unsloth/Qwen2-1.5B-Instruct-bnb-4bit" : (
"unsloth/Qwen2-1.5B-Instruct",
"Qwen/Qwen2-1.5B-Instruct",
),
"unsloth/Qwen2-7B-bnb-4bit" : (
"unsloth/Qwen2-7B",
"Qwen/Qwen2-7B",
),
"unsloth/Qwen2-7B-Instruct-bnb-4bit" : (
"unsloth/Qwen2-7B-Instruct",
"Qwen/Qwen2-7B-Instruct",
),
"unsloth/Qwen2-70B-bnb-4bit" : (
"Qwen/Qwen2-70B",
),
"unsloth/Qwen2-70B-Instruct-bnb-4bit" : (
"Qwen/Qwen2-70B-Instruct",
),
}
INT_TO_FLOAT_MAPPER = {}