add rlhf-v dataset

Former-commit-id: 8e49940746c1a6ff910f07dbefbec14af9d0f3c6

README.md

@@ -291,6 +291,7 @@ You also can add a custom chat template to [template.py](src/llamafactory/data/t
 
 - [DPO mixed (en&zh)](https://huggingface.co/datasets/hiyouga/DPO-En-Zh-20k)
 - [UltraFeedback (en)](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized)
+- [RLHF-V (en)](https://huggingface.co/datasets/openbmb/RLHF-V-Dataset)
 - [Orca DPO Pairs (en)](https://huggingface.co/datasets/Intel/orca_dpo_pairs)
 - [HH-RLHF (en)](https://huggingface.co/datasets/Anthropic/hh-rlhf)
 - [Nectar (en)](https://huggingface.co/datasets/berkeley-nest/Nectar)

README_zh.md

@@ -292,6 +292,7 @@ https://github.com/user-attachments/assets/e6ce34b0-52d5-4f3e-a830-592106c4c272
 
 - [DPO mixed (en&zh)](https://huggingface.co/datasets/hiyouga/DPO-En-Zh-20k)
 - [UltraFeedback (en)](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized)
+- [RLHF-V (en)](https://huggingface.co/datasets/openbmb/RLHF-V-Dataset)
 - [Orca DPO Pairs (en)](https://huggingface.co/datasets/Intel/orca_dpo_pairs)
 - [HH-RLHF (en)](https://huggingface.co/datasets/Anthropic/hh-rlhf)
 - [Nectar (en)](https://huggingface.co/datasets/berkeley-nest/Nectar)

data/dataset_info.json

@@ -433,6 +433,17 @@
       "rejected": "rejected"
     }
   },
+  "rlhf_v": {
+    "hf_hub_url": "llamafactory/RLHF-V",
+    "ranking": true,
+    "formatting": "sharegpt",
+    "columns": {
+      "messages": "conversations",
+      "chosen": "chosen",
+      "rejected": "rejected",
+      "images": "images"
+    }
+  },
   "orca_pairs": {
     "hf_hub_url": "Intel/orca_dpo_pairs",
     "ranking": true,

examples/README.md

@@ -36,6 +36,18 @@ llamafactory-cli train examples/train_lora/llava1_5_lora_sft.yaml
 llamafactory-cli train examples/train_lora/qwen2vl_lora_sft.yaml
 ```
 
+#### DPO/ORPO/SimPO Training
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_dpo.yaml
+```
+
+#### Multimodal DPO/ORPO/SimPO Training
+
+```bash
+llamafactory-cli train examples/train_lora/qwen2vl_lora_dpo.yaml
+```
+
 #### Reward Modeling
 
 ```bash
@@ -48,12 +60,6 @@ llamafactory-cli train examples/train_lora/llama3_lora_reward.yaml
 llamafactory-cli train examples/train_lora/llama3_lora_ppo.yaml
 ```
 
-#### DPO/ORPO/SimPO Training
-
-```bash
-llamafactory-cli train examples/train_lora/llama3_lora_dpo.yaml
-```
-
 #### KTO Training
 
 ```bash

examples/README_zh.md

@@ -36,6 +36,18 @@ llamafactory-cli train examples/train_lora/llava1_5_lora_sft.yaml
 llamafactory-cli train examples/train_lora/qwen2vl_lora_sft.yaml
 ```
 
+#### DPO/ORPO/SimPO 训练
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_dpo.yaml
+```
+
+#### 多模态 DPO/ORPO/SimPO 训练
+
+```bash
+llamafactory-cli train examples/train_lora/qwen2vl_lora_dpo.yaml
+```
+
 #### 奖励模型训练
 
 ```bash
@@ -48,12 +60,6 @@ llamafactory-cli train examples/train_lora/llama3_lora_reward.yaml
 llamafactory-cli train examples/train_lora/llama3_lora_ppo.yaml
 ```
 
-#### DPO/ORPO/SimPO 训练
-
-```bash
-llamafactory-cli train examples/train_lora/llama3_lora_dpo.yaml
-```
-
 #### KTO 训练
 
 ```bash

examples/train_lora/qwen2vl_lora_dpo.yaml

@@ -0,0 +1,41 @@
+### model
+model_name_or_path: Qwen/Qwen2-VL-7B-Instruct
+
+### method
+stage: dpo
+do_train: true
+finetuning_type: lora
+lora_target: all
+pref_beta: 0.1
+pref_loss: sigmoid  # choices: [sigmoid (dpo), orpo, simpo]
+
+### dataset
+dataset: rlhf_v
+template: qwen2_vl
+cutoff_len: 1024
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+
+### output
+output_dir: saves/qwen2_vl-7b/lora/dpo
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 5.0e-6
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+val_size: 0.1
+per_device_eval_batch_size: 1
+eval_strategy: steps
+eval_steps: 500

src/llamafactory/data/aligner.py

@@ -14,7 +14,7 @@
 
 import os
 from functools import partial
-from typing import TYPE_CHECKING, Any, Dict, List, Union
+from typing import TYPE_CHECKING, Any, Dict, List, Sequence, Union
 
 from datasets import Features
 
@@ -33,19 +33,17 @@ if TYPE_CHECKING:
 logger = get_logger(__name__)
 
 
-def _convert_images(images: List[Any], dataset_attr: "DatasetAttr", data_args: "DataArguments") -> List[Any]:
+def _convert_images(images: Sequence[Any], dataset_attr: "DatasetAttr", data_args: "DataArguments") -> List[Any]:
     r"""
     Optionally concatenates image path to dataset dir when loading from local disk.
     """
-    outputs = []
+    images = images[:]
     if dataset_attr.load_from in ["script", "file"]:
-        for image in images:
+        for i in range(len(images)):
-            if isinstance(image, str) and os.path.isfile(os.path.join(data_args.dataset_dir, image)):
+            if isinstance(images[i], str) and os.path.isfile(os.path.join(data_args.dataset_dir, images[i])):
-                outputs.append(os.path.join(data_args.dataset_dir, image))
+                images[i] = os.path.join(data_args.dataset_dir, images[i])
-            else:
-                outputs.append(image)
 
-    return outputs
+    return images
 
 
 def convert_alpaca(

src/llamafactory/data/collator.py

@@ -142,15 +142,15 @@ class PairwiseDataCollatorWithPadding(MultiModalDataCollatorForSeq2Seq):
                     "attention_mask": feature["{}_attention_mask".format(key)],
                     "labels": feature["{}_labels".format(key)],
                 }
+                if "{}_token_type_ids".format(key) in feature:
+                    target_feature["token_type_ids"] = feature["{}_token_type_ids".format(key)]
+
                 if "pixel_values" in feature:  # image data are same for chosen and rejected
                     target_feature["pixel_values"] = feature["pixel_values"]
 
                 if "image_grid_thw" in feature:
                     target_feature["image_grid_thw"] = feature["image_grid_thw"]
 
-                if "{}_token_type_ids".format(key) in feature:
-                    target_feature["token_type_ids"] = feature["{}_token_type_ids".format(key)]
-
                 concatenated_features.append(target_feature)
 
         return super().__call__(concatenated_features)
@@ -177,16 +177,16 @@ class KTODataCollatorWithPadding(MultiModalDataCollatorForSeq2Seq):
                 "attention_mask": feature["kl_attention_mask"],
                 "labels": feature["kl_labels"],
             }
+            if "token_type_ids" in feature:
+                target_feature["token_type_ids"] = feature["token_type_ids"]
+                kl_feature["token_type_ids"] = feature["kl_token_type_ids"]
+
             if "pixel_values" in feature:
                 target_feature["pixel_values"] = feature["pixel_values"]
 
             if "image_grid_thw" in feature:
                 target_feature["image_grid_thw"] = feature["image_grid_thw"]
 
-            if "token_type_ids" in feature:
-                target_feature["token_type_ids"] = feature["token_type_ids"]
-                kl_feature["token_type_ids"] = feature["kl_token_type_ids"]
-
             target_features.append(target_feature)
             kl_features.append(kl_feature)
             kto_tags.append(feature["kto_tags"])

src/llamafactory/data/mm_plugin.py

@@ -19,6 +19,23 @@ if TYPE_CHECKING:
     from transformers.image_processing_utils import BaseImageProcessor
 
 
+def _regularize_images(images: Sequence["ImageObject"], processor: "ProcessorMixin") -> List["ImageObject"]:
+    r"""
+    Regularizes images to avoid error. Including resizing and mode convert.
+    """
+    images = images[:]
+    image_resolution = getattr(processor, "image_resolution", 512)
+    for i in range(len(images)):
+        if max(images[i].width, images[i].height) > image_resolution:
+            factor = image_resolution / max(images[i].width, images[i].height)
+            images[i] = images[i].resize((int(images[i].width * factor), int(images[i].height * factor)))
+
+        if images[i].mode != "RGB":
+            images[i] = images[i].convert("RGB")
+
+    return images
+
+
 def _get_mm_inputs(images: Sequence["ImageObject"], processor: "ProcessorMixin") -> Dict[str, "torch.Tensor"]:
     r"""
     Processes visual inputs.
@@ -34,6 +51,7 @@ def _get_mm_inputs(images: Sequence["ImageObject"], processor: "ProcessorMixin")
     """
     image_processor: "BaseImageProcessor" = getattr(processor, "image_processor")
     if len(images) != 0:
+        images = _regularize_images(images, processor)
         image_inputs = image_processor(images=images, return_tensors="pt")
     else:  # add NoneType for fake images
         image = Image.new("RGB", (64, 64), (255, 255, 255))

src/llamafactory/hparams/model_args.py

@@ -138,6 +138,10 @@ class ModelArguments:
         default=False,
         metadata={"help": "Whether or not to randomly initialize the model weights."},
     )
+    image_resolution: int = field(
+        default=512,
+        metadata={"help": "Keeps the height or width of image below this resolution."},
+    )
     infer_backend: Literal["huggingface", "vllm"] = field(
         default="huggingface",
         metadata={"help": "Backend engine used at inference."},

src/llamafactory/model/loader.py

@@ -99,6 +99,7 @@ def load_tokenizer(model_args: "ModelArguments") -> "TokenizerModule":
         processor = AutoProcessor.from_pretrained(model_args.model_name_or_path, **init_kwargs)
         setattr(processor, "tokenizer", tokenizer)
         setattr(processor, "image_seqlen", get_image_seqlen(config))
+        setattr(processor, "image_resolution", model_args.image_resolution)
     except Exception:
         processor = None
 

src/llamafactory/train/dpo/trainer.py

@@ -176,7 +176,6 @@ class CustomDPOTrainer(DPOTrainer):
             batch = {k: v.detach().clone() for k, v in batch.items()}  # avoid error
 
         all_logits: "torch.Tensor" = model(**batch, return_dict=True, use_cache=False).logits.to(torch.float32)
-
         all_logps, valid_length = get_batch_logps(logits=all_logits, labels=batch["labels"])
         if self.loss_type in ["ipo", "orpo", "simpo"]:
             all_logps = all_logps / valid_length

src/llamafactory/train/kto/trainer.py

@@ -127,17 +127,16 @@ class CustomKTOTrainer(KTOTrainer):
             "input_ids": batch["{}input_ids".format(prefix)],
             "attention_mask": batch["{}attention_mask".format(prefix)],
         }
+        if "{}token_type_ids".format(prefix) in batch:
+            model_inputs["token_type_ids"] = batch["{}token_type_ids".format(prefix)]
+
         if "pixel_values" in batch:
             model_inputs["pixel_values"] = batch["pixel_values"]
 
         if "image_grid_thw" in batch:
             model_inputs["image_grid_thw"] = batch["image_grid_thw"]
 
-        if "{}token_type_ids".format(prefix) in batch:
-            model_inputs["token_type_ids"] = batch["{}token_type_ids".format(prefix)]
-
         logits = model(**model_inputs, return_dict=True, use_cache=False).logits.to(torch.float32)
-
         logps, valid_length = get_batch_logps(logits=logits, labels=batch["{}labels".format(prefix)])
         return logps, logps / valid_length