fix trl, release

Author: abhishekkrthakur

configs/llm_finetuning/smollm2_orpo.yml

@@ -0,0 +1,36 @@
+task: llm-orpo
+base_model: HuggingFaceTB/SmolLM2-1.7B-Instruct
+project_name: autotrain-smallm2-orpo
+log: tensorboard
+backend: local
+
+data:
+  path: argilla/distilabel-capybara-dpo-7k-binarized
+  train_split: train
+  valid_split: null
+  chat_template: chatml
+  column_mapping:
+    text_column: chosen
+    rejected_text_column: rejected
+    prompt_text_column: prompt
+
+params:
+  block_size: 1024
+  model_max_length: 2048
+  max_prompt_length: 512
+  epochs: 3
+  batch_size: 2
+  lr: 3e-5
+  peft: true
+  quantization: int4
+  target_modules: all-linear
+  padding: right
+  optimizer: adamw_torch
+  scheduler: linear
+  gradient_accumulation: 4
+  mixed_precision: fp16
+
+hub:
+  username: ${HF_USERNAME}
+  token: ${HF_TOKEN}
+  push_to_hub: false

src/autotrain/__init__.py

@@ -45,7 +45,7 @@
 warnings.filterwarnings("ignore", category=UserWarning, module="huggingface_hub")
 
 logger = Logger().get_logger()
-__version__ = "0.8.30.dev0"
+__version__ = "0.8.30"
 
 
 def is_colab():

src/autotrain/trainers/clm/train_clm_dpo.py

@@ -109,7 +109,7 @@ def train(config):
         ref_model=model_ref,
         train_dataset=train_data,
         eval_dataset=valid_data if config.valid_split is not None else None,
-        tokenizer=tokenizer,
+        processing_class=tokenizer,
         peft_config=peft_config if config.peft else None,
     )
 

src/autotrain/trainers/clm/train_clm_orpo.py

@@ -48,7 +48,7 @@ def train(config):
         **trainer_args,
         train_dataset=train_data,
         eval_dataset=valid_data if config.valid_split is not None else None,
-        tokenizer=tokenizer,
+        processing_class=tokenizer,
         peft_config=peft_config if config.peft else None,
     )
 

src/autotrain/trainers/clm/train_clm_reward.py

@@ -116,7 +116,7 @@ def train(config):
         train_dataset=train_data,
         eval_dataset=valid_data if config.valid_split is not None else None,
         peft_config=peft_config if config.peft else None,
-        tokenizer=tokenizer,
+        processing_class=tokenizer,
     )
 
     trainer.remove_callback(PrinterCallback)

src/autotrain/trainers/clm/train_clm_sft.py

@@ -48,7 +48,7 @@ def train(config):
         train_dataset=train_data,
         eval_dataset=valid_data if config.valid_split is not None else None,
         peft_config=peft_config if config.peft else None,
-        tokenizer=tokenizer,
+        processing_class=tokenizer,
     )
 
     trainer.remove_callback(PrinterCallback)