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

Fix renaming on other model than Llama (#2762)

Browse source
This commit is contained in:
Edd 2025-06-19 04:38:36 +08:00 • committed by GitHub
commit 3a7cf48532
8 changed files with 8 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

2
unsloth/models/cohere.py
View file

@ -466,7 +466,7 @@ class FastCohereModel(FastLlamaModel):
CohereDecoderLayer .forward = CohereDecoderLayer_fast_forward
CohereModel .forward = LlamaModel_fast_forward
CohereForCausalLM .forward = CausalLM_fast_forward(CohereModel_fast_forward_inference)
PeftModelForCausalLM .forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM .forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(CohereForCausalLM)
import transformers.models.cohere.modeling_cohere

2
unsloth/models/gemma.py
View file

@ -332,7 +332,7 @@ class FastGemmaModel(FastLlamaModel):
GemmaDecoderLayer .forward = GemmaDecoderLayer_fast_forward
GemmaModel .forward = LlamaModel_fast_forward
GemmaForCausalLM .forward = CausalLM_fast_forward(GemmaModel_fast_forward_inference)
PeftModelForCausalLM.forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM.forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(GemmaForCausalLM)
# Solves https://github.com/unslothai/unsloth/issues/168

2
unsloth/models/gemma2.py
View file

@ -477,7 +477,7 @@ class FastGemma2Model(FastLlamaModel):
Gemma2DecoderLayer .forward = Gemma2DecoderLayer_fast_forward
Gemma2Model .forward = LlamaModel_fast_forward
Gemma2ForCausalLM .forward = CausalLM_fast_forward(Gemma2Model_fast_forward_inference)
PeftModelForCausalLM .forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM .forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(Gemma2ForCausalLM)
# Solves https://github.com/unslothai/unsloth/issues/168

2
unsloth/models/granite.py
View file

@ -468,7 +468,7 @@ class FastGraniteModel(FastLlamaModel):
GraniteModel .forward = LlamaModel_fast_forward
GraniteForCausalLM .forward = CausalLM_fast_forward(GraniteModel_fast_forward_inference)
GraniteForCausalLM .__init__ = patched_init(GraniteForCausalLM.__init__)
PeftModelForCausalLM .forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM .forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(GraniteForCausalLM)
import transformers.models.granite.modeling_granite

2
unsloth/models/mistral.py
View file

@ -368,7 +368,7 @@ class FastMistralModel(FastLlamaModel):
MistralDecoderLayer .forward = LlamaDecoderLayer_fast_forward
MistralModel .forward = LlamaModel_fast_forward
MistralForCausalLM .forward = MistralForCausalLM_fast_forward
PeftModelForCausalLM .forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM .forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(MistralForCausalLM)
# Solves https://github.com/unslothai/unsloth/issues/168

2
unsloth/models/qwen2.py
View file

@ -55,7 +55,7 @@ class FastQwen2Model(FastLlamaModel):
Qwen2DecoderLayer .forward = LlamaDecoderLayer_fast_forward
Qwen2Model .forward = LlamaModel_fast_forward
Qwen2ForCausalLM .forward = CausalLM_fast_forward(LlamaModel_fast_forward_inference)
PeftModelForCausalLM.forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM.forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(Qwen2ForCausalLM)
# Solves https://github.com/unslothai/unsloth/issues/168

2
unsloth/models/qwen3.py
View file

@ -387,7 +387,7 @@ class FastQwen3Model(FastLlamaModel):
Qwen3DecoderLayer .forward = LlamaDecoderLayer_fast_forward
Qwen3Model .forward = LlamaModel_fast_forward
Qwen3ForCausalLM .forward = CausalLM_fast_forward(_LlamaModel_fast_forward_inference(Qwen3Attention_fast_forward_inference))
PeftModelForCausalLM.forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM.forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(Qwen3ForCausalLM)
# Solves https://github.com/unslothai/unsloth/issues/168

2
unsloth/models/qwen3_moe.py
View file

@ -177,7 +177,7 @@ class FastQwen3MoeModel(FastQwen3Model):
Qwen3MoeDecoderLayer .forward = Qwen3MoeDecoderLayer_fast_forward
Qwen3MoeModel .forward = LlamaModel_fast_forward
Qwen3MoeForCausalLM .forward = CausalLM_fast_forward(LlamaModel_fast_forward_inference)
PeftModelForCausalLM.forward = PeftModelForCausalLM_fast_forward
PeftModelForCausalLM.forward = PeftModel_fast_forward
fix_prepare_inputs_for_generation(Qwen3MoeForCausalLM)
# Solves https://github.com/unslothai/unsloth/issues/168