🗜 Hotfix: avoid passing quantization_config=None

examples/scripts/dpo_vlm.py

@@ -88,15 +88,18 @@
     # Model & Tokenizer
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
 
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     model = AutoModelForImageTextToText.from_pretrained(
         model_args.model_name_or_path,
         trust_remote_code=model_args.trust_remote_code,

examples/scripts/gkd.py

@@ -81,16 +81,19 @@
     ################
     # Model & Tokenizer
     ################
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         trust_remote_code=model_args.trust_remote_code,
         attn_implementation=model_args.attn_implementation,
         dtype=model_args.dtype,
         use_cache=False if training_args.gradient_checkpointing else True,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     training_args.model_init_kwargs = model_kwargs
 
     teacher_model_kwargs = dict(
@@ -99,9 +102,12 @@
         attn_implementation=model_args.attn_implementation,
         dtype=model_args.dtype,
         use_cache=True,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     training_args.teacher_model_init_kwargs = teacher_model_kwargs
 
     tokenizer = AutoTokenizer.from_pretrained(

examples/scripts/grpo_vlm.py

@@ -97,14 +97,16 @@
     # Model & Processor
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     training_args.model_init_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        training_args.model_init_kwargs["device_map"] = get_kbit_device_map()
+        training_args.model_init_kwargs["quantization_config"] = quantization_config
 
     ################
     # Dataset

examples/scripts/gspo.py

@@ -83,14 +83,16 @@
     # Model & Processor
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     training_args.model_init_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        training_args.model_init_kwargs["device_map"] = get_kbit_device_map()
+        training_args.model_init_kwargs["quantization_config"] = quantization_config
 
     ################
     # Dataset

examples/scripts/gspo_vlm.py

@@ -84,14 +84,16 @@
     # Model & Processor
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     training_args.model_init_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        training_args.model_init_kwargs["device_map"] = get_kbit_device_map()
+        training_args.model_init_kwargs["quantization_config"] = quantization_config
 
     ################
     # Dataset

examples/scripts/mpo_vlm.py

@@ -72,16 +72,19 @@
     # Model & Processor
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
 
     model_kwargs = dict(
         trust_remote_code=model_args.trust_remote_code,
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     model = AutoModelForImageTextToText.from_pretrained(
         model_args.model_name_or_path,
         **model_kwargs,

examples/scripts/nash_md.py

@@ -88,15 +88,17 @@
     training_args.gradient_checkpointing_kwargs = {"use_reentrant": True}
 
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
         use_cache=False if training_args.gradient_checkpointing else True,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     model = AutoModelForCausalLM.from_pretrained(
         model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, **model_kwargs

examples/scripts/online_dpo.py

@@ -84,15 +84,17 @@
     training_args.gradient_checkpointing_kwargs = {"use_reentrant": True}
 
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
         use_cache=False if training_args.gradient_checkpointing else True,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     model = AutoModelForCausalLM.from_pretrained(
         model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, **model_kwargs

examples/scripts/online_dpo_vlm.py

@@ -115,15 +115,17 @@
     training_args.gradient_checkpointing_kwargs = {"use_reentrant": True}
 
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
         use_cache=False if training_args.gradient_checkpointing else True,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     # Load the VLM model using correct architecture (from GRPO pattern)
     config = AutoConfig.from_pretrained(model_args.model_name_or_path)

examples/scripts/ppo/ppo.py

@@ -91,14 +91,16 @@
     # Model & Tokenizer
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     tokenizer = AutoTokenizer.from_pretrained(
         model_args.model_name_or_path, padding_side="left", trust_remote_code=model_args.trust_remote_code

examples/scripts/ppo/ppo_tldr.py

@@ -98,14 +98,16 @@
     # Model & Tokenizer
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     tokenizer = AutoTokenizer.from_pretrained(
         model_args.model_name_or_path, padding_side="left", trust_remote_code=model_args.trust_remote_code

examples/scripts/prm.py

@@ -76,30 +76,33 @@
 
 if __name__ == "__main__":
     parser = HfArgumentParser((ScriptArguments, PRMConfig, ModelConfig))
-    script_args, training_args, model_config = parser.parse_args_into_dataclasses()
+    script_args, training_args, model_args = parser.parse_args_into_dataclasses()
     training_args.gradient_checkpointing_kwargs = dict(use_reentrant=False)
 
     ################
     # Model & Tokenizer
     ################
-    dtype = model_config.dtype if model_config.dtype in ["auto", None] else getattr(torch, model_config.dtype)
-    quantization_config = get_quantization_config(model_config)
+    dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
     model_kwargs = dict(
-        revision=model_config.model_revision,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
+        revision=model_args.model_revision,
         use_cache=False if training_args.gradient_checkpointing else True,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     tokenizer = AutoTokenizer.from_pretrained(
-        model_config.model_name_or_path, trust_remote_code=model_config.trust_remote_code, use_fast=True
+        model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, use_fast=True
     )
     model = AutoModelForTokenClassification.from_pretrained(
-        model_config.model_name_or_path, num_labels=2, trust_remote_code=model_config.trust_remote_code, **model_kwargs
+        model_args.model_name_or_path, num_labels=2, trust_remote_code=model_args.trust_remote_code, **model_kwargs
     )
     # Align padding tokens between tokenizer and model
     model.config.pad_token_id = tokenizer.pad_token_id
 
-    if model_config.use_peft and model_config.lora_task_type != "TOKEN_CLS":
+    if model_args.use_peft and model_args.lora_task_type != "TOKEN_CLS":
         logger.warning(
             "You are using a `task_type` that is different than `TOKEN_CLS` for PEFT. This will lead to silent bugs"
             " Make sure to pass --lora_task_type TOKEN_CLS when using this script with PEFT.",
@@ -121,7 +124,7 @@
         args=training_args,
         train_dataset=dataset[script_args.dataset_train_split],
         eval_dataset=dataset[script_args.dataset_test_split],
-        peft_config=get_peft_config(model_config),
+        peft_config=get_peft_config(model_args),
     )
     trainer.train()
 

examples/scripts/reward_modeling.py

@@ -86,14 +86,17 @@
     # Model & Tokenizer
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
         use_cache=False if training_args.gradient_checkpointing else True,
         dtype=dtype,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     tokenizer = AutoTokenizer.from_pretrained(
         model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, use_fast=True
     )

examples/scripts/sft_gpt_oss.py

@@ -31,7 +31,7 @@
     examples/scripts/sft_gpt_oss.py \
     --dtype bfloat16 \
     --model_name_or_path openai/gpt-oss-20b \
-    --packing true packing_strategy wrapped \
+    --packing \
     --run_name 20b-full-eager \
     --attn_implementation kernels-community/vllm-flash-attn3 \
     --dataset_num_proc 12 \

examples/scripts/sft_vlm.py

@@ -85,14 +85,16 @@
     # Model, Tokenizer & Processor
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     model = AutoModelForImageTextToText.from_pretrained(
         model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, **model_kwargs

examples/scripts/sft_vlm_gemma3.py

@@ -150,14 +150,17 @@ def main():
     # Model, Tokenizer & Processor
     ################
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
+
     model = AutoModelForImageTextToText.from_pretrained(
         model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, **model_kwargs
     )

examples/scripts/xpo.py

@@ -73,15 +73,17 @@
     training_args.gradient_checkpointing_kwargs = {"use_reentrant": True}
 
     dtype = model_args.dtype if model_args.dtype in ["auto", None] else getattr(torch, model_args.dtype)
-    quantization_config = get_quantization_config(model_args)
     model_kwargs = dict(
         revision=model_args.model_revision,
         attn_implementation=model_args.attn_implementation,
         dtype=dtype,
         use_cache=False if training_args.gradient_checkpointing else True,
-        device_map=get_kbit_device_map() if quantization_config is not None else None,
-        quantization_config=quantization_config,
     )
+    quantization_config = get_quantization_config(model_args)
+    if quantization_config is not None:
+        # Passing None would not be treated the same as omitting the argument, so we include it only when valid.
+        model_kwargs["device_map"] = get_kbit_device_map()
+        model_kwargs["quantization_config"] = quantization_config
 
     model = AutoModelForCausalLM.from_pretrained(
         model_args.model_name_or_path, trust_remote_code=model_args.trust_remote_code, **model_kwargs