Update README.md

README.md

@@ -3,199 +3,146 @@ base_model: OpenGVLab/InternVL2-4B
 library_name: peft
 ---
 
-# Model Card for Model ID
-<!-- Provide a quick summary of what the model is/does. -->
+# Model Details
 
+- **Developed by:** Jian Chen
+- **Model type:** MLLM-based encoder
+- **Finetuned from model:** [OpenGVLab/InternVL2-4B](https://huggingface.co/OpenGVLab/InternVL2-4B)
 
-
-## Model Details
-
-### Model Description
-
-<!-- Provide a longer summary of what this model is. -->
-
-
-
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
-
-### Model Sources [optional]
+## Model Sources [optional]
 
 <!-- Provide the basic links for the model. -->
 
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+- **Repository:** [SV-RAG](https://github.com/puar-playground/SV-RAG)
+- **Paper [optional]:** [SV-RAG: LoRA-Contextualizing Adaptation of Large Multimodal Models for Long Document Understanding](https://arxiv.org/abs/2411.01106)
 
 ## Uses
-
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
-
-### Direct Use
-
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
-
-[More Information Needed]
-
-### Downstream Use [optional]
-
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
-
-[More Information Needed]
-
-### Out-of-Scope Use
-
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-
-[More Information Needed]
-
-## Bias, Risks, and Limitations
-
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-
-[More Information Needed]
-
-### Recommendations
-
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
-
-## How to Get Started with the Model
-
-Use the code below to get started with the model.
-
-[More Information Needed]
-
-## Training Details
-
-### Training Data
-
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
-
-[More Information Needed]
-
-### Training Procedure
-
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-
-#### Preprocessing [optional]
-
-[More Information Needed]
-
-
-#### Training Hyperparameters
-
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-
-#### Speeds, Sizes, Times [optional]
-
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-
-[More Information Needed]
-
-## Evaluation
-
-<!-- This section describes the evaluation protocols and provides the results. -->
-
-### Testing Data, Factors & Metrics
-
-#### Testing Data
-
-<!-- This should link to a Dataset Card if possible. -->
-
-[More Information Needed]
-
-#### Factors
-
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-
-[More Information Needed]
-
-#### Metrics
-
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-
-[More Information Needed]
-
-### Results
-
-[More Information Needed]
-
-#### Summary
-
-
-
-## Model Examination [optional]
-
-<!-- Relevant interpretability work for the model goes here -->
-
-[More Information Needed]
-
-## Environmental Impact
-
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
-
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
-
-## Technical Specifications [optional]
-
-### Model Architecture and Objective
-
-[More Information Needed]
-
-### Compute Infrastructure
-
-[More Information Needed]
-
-#### Hardware
-
-[More Information Needed]
-
-#### Software
-
-[More Information Needed]
+A demo script is provided in the [GitHub](https://github.com/puar-playground/SV-RAG/blob/main/test_retrieval.py)
+
+Alternatively, this code provides a more detailed breakdown of the computation. The [`colpali_engine`](https://github.com/puar-playground/SV-RAG/tree/main/colpali_engine) used is customized and is available in the GitHub.
+```
+from colpali_engine.models import ColInternvl2_4b, ColInternProcessor
+
+class ColInternVL2Retriever(BaseRetriever):
+    """Retriever class using ColInternVL2 for multimodal retrieval."""
+
+    def __init__(self, model_name="puar-playground/Col-InternVL2-4B", device="cuda" if torch.cuda.is_available() else "cpu"):
+        """
+        Initializes the ColInternVL2 model.
+
+        Args:
+            model_name (str): The model identifier.
+            device (str): Device to run the model on ('cuda' or 'cpu').
+        """
+        os.system('pip install transformers==4.47.1')
+        self.multimodel = True
+        self.device = device
+
+        self.model = ColInternvl2_4b.from_pretrained(
+            model_name,
+            torch_dtype=torch.bfloat16,
+            device_map=device).eval()
+
+        self.processor = ColInternProcessor('OpenGVLab/InternVL2-4B')
+
+    def process_text(self, query_list: List[str], batch_size: int = 4):
+        """
+        Processes a list of text queries into embeddings using ColPhi in batches.
+
+        Args:
+            query_list (List[str]): List of query texts.
+            batch_size (int): Number of queries processed per batch.
+
+        Returns:
+            torch.Tensor: Concatenated embeddings for all queries.
+        """
+        all_embeddings = []
+
+        for i in range(0, len(query_list), batch_size):
+            batch_queries = query_list[i : i + batch_size]
+
+            # Convert queries to model-compatible format
+            batch_inputs = self.processor.process_queries(batch_queries).to(self.model.device)
+
+            with torch.no_grad():
+                batch_embeddings = self.model(**batch_inputs)
+
+            all_embeddings.append(batch_embeddings.to("cpu"))
+        
+        # Concatenate all batch outputs into a single tensor
+        all_embeddings = self.pad_and_cat_tensors(all_embeddings)
+
+        return all_embeddings
+    
+    @staticmethod
+    def pad_and_cat_tensors(tensor_list):
+        # Find the maximum length of the second dimension (x_i) across all tensors
+        max_x = max(tensor.size(1) for tensor in tensor_list)
+        
+        # Pad tensors to have the same size in the second dimension
+        padded_tensors = []
+        for tensor in tensor_list:
+            padding_size = max_x - tensor.size(1)
+            # Pad with zeros on the right in the second dimension
+            padded_tensor = torch.nn.functional.pad(tensor, (0, 0, 0, padding_size))
+            padded_tensors.append(padded_tensor)
+        
+        # Concatenate the padded tensors along the first dimension
+        result_tensor = torch.cat(padded_tensors, dim=0)
+        
+        return result_tensor
+
+    def process_image(self, image_dir_list: List[str]):
+        """Processes images into embeddings using ColInternVL2."""
+        def process_images_in_batches(processor, img_dir_list, model, batch_size=2):
+            all_embeddings = []
+            
+            # Split img_dir_list into batches
+            for img_dir in img_dir_list:
+
+                img = Image.open(img_dir)
+        
+                # Process the batch of images
+                batch_features = processor.process_images(img)
+                
+                # Extract the tensor from the BatchFeature object
+                batch_images = {k: v.to(model.device) for k, v in batch_features.items()}
+        
+                # Assuming the model expects a specific input (e.g., 'pixel_values')
+                embeddings = model(**batch_images)
+                
+                # Move embeddings to CPU and append to the list
+                embeddings = embeddings.to("cpu")
+                all_embeddings.append(embeddings)
+
+            # Concatenate all processed batches into a single tensor
+            all_embeddings = self.pad_and_cat_tensors(all_embeddings)
+            return all_embeddings
+        
+        # Forward pass
+        with torch.no_grad():
+            # image_embeddings = model(**batch_images)
+            image_embeddings = process_images_in_batches(self.processor, image_dir_list, self.model)
+                    
+        return image_embeddings
+
+    def compute_similarity(self, text_embeddings, image_embeddings):
+        """ Computes cosine similarity between text and image embeddings. """
+        scores = self.processor.score_multi_vector(text_embeddings, image_embeddings)
+        return scores
+
+    def retrieve(self, query_list: str, image_list: List[str]):
+
+        text_embeddings = self.process_text(query_list)
+        image_embeddings = self.process_image(image_list)
+
+        similarity_score = self.compute_similarity(text_embeddings, image_embeddings)
+        values, top_indices = torch.tensor(similarity_score).sort(descending=True)
+
+        return values, top_indices
+```
 
 ## Citation [optional]
 
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
-
-**BibTeX:**
-
-[More Information Needed]
-
-**APA:**
-
-[More Information Needed]
-
-## Glossary [optional]
-
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
-
-[More Information Needed]
-
-## More Information [optional]
-
-[More Information Needed]
-
-## Model Card Authors [optional]
-
-[More Information Needed]
-
-## Model Card Contact
 
-[More Information Needed]
-### Framework versions
 
-- PEFT 0.11.1