Update README.md

README.md

@@ -1,13 +1,17 @@
----
-library_name: transformers
-tags: []
----
+---
+library_name: transformers
+license: apache-2.0
+language:
+- en
+base_model:
+- sentence-transformers/all-distilroberta-v1
+---
 
-# Model Card for Model ID
+# distilroberta-ce-esci
 
 <!-- Provide a quick summary of what the model is/does. -->
 
-
+This is a cross-encoder model optimized for e-commerce text classification tasks.
 
 ## Model Details
 
@@ -15,23 +19,20 @@ tags: []
 
 <!-- Provide a longer summary of what this model is. -->
 
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
+This is a fine-tuned cross-encoder model based on all-distilroberta-v1, trained on an [e-commerce dataset](https://github.com/amazon-science/esci-data/tree/main/shopping_queries_dataset) of query-product pairs. The model predicts relevance classes in the ESCI (Exact, Substitute, Complementary, Irrelevant) framework by capturing the relationship of the input text and class labels, which can be used for multi-class classification tasks or more complex downstream tasks.
 
-- **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]
+- **Developed by:** Sarah Lawlis / DASC Practicum Team 12
+- **Shared by:** University of Arkansas Data Science Practicum Team 12
+- **Model type:** Sequence Classification (Cross-Encoder)
+- **Language(s) (NLP):** English
+- **License:** apache-2.0
+- **Finetuned from model:** sentence-transformers/all-distilroberta-v1
 
-### Model Sources [optional]
+### Model Sources
 
 <!-- Provide the basic links for the model. -->
 
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+- **Repository:** [sllawlis/distilroberta-ce-esci](https://huggingface.co/sllawlis/distilroberta-ce-esci)
 
 ## Uses
 
@@ -41,159 +42,92 @@ This is the model card of a 🤗 transformers model that has been pushed on the
 
 <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
 
-[More Information Needed]
+This model is designed for multi-class product classification within the ESCI framework. The model directly predicts one of the ESCI labels for a given query-product pair. This task is the foundation for downstream use cases.
 
-### Downstream Use [optional]
+### Downstream Use
 
 <!-- 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
+The model's multi-class predictions can be used in the following downstream tasks:
+1. Ranking Systems:
+    * Combine the model's predictions with bi-encoders for a two-stage ranking pipeline:
+      * First Stage (Bi-Encoders): Generate candidate products efficiently by retrieving embeddings of query and product titles
+      * Second Stage (Cross-Encoders): Re-rank the candidates using fine-grained ESCI label predictions for better accuracy
 
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-
-[More Information Needed]
+2. Product Substitute Identification:
+    * Use the Substitute label from the model to identify products that can replace one another
 
 ## 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.
+* Bias: Due to heavy imbalance in ESCI labels in the training data, this model's predictions may skew to predicting more Exact labels.
+* Limitations: This model is domain-specific to e-commerce data and may not generalize well to other domains. This model is optimized for the English language and may perform poorly with non-English data. Cross-encoders are computationally expensive for large-scale applications, there may be difficulty implementing this model for real-time inference.
 
 ## 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
+```python
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
 
-<!-- This should link to a Dataset Card if possible. -->
+# Load the tokenizer and model
+model_name = "sllawlis/distilroberta-ce-esci"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSequenceClassification.from_pretrained(model_name)
+```
 
-[More Information Needed]
+## Usage (Multi-class Classification Example)
 
-#### Factors
+Below is a quick usage example of this model.
 
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+```python
+# Example query-product pair
+query = "wireless headphones"
+product = "Noise-cancelling wireless headphones with long battery life"
 
-[More Information Needed]
+# Tokenize inputs
+inputs = tokenizer(
+    query,
+    product,
+    truncation=True,
+    padding=True,
+    return_tensors="pt"
+)
 
-#### Metrics
+# Predict relevance
+outputs = model(**inputs)
+predicted_class = outputs.logits.argmax(dim=1).item()
+print(f"Predicted Class: {predicted_class}")
+```
 
-<!-- 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]
-
-## 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. -->
+## Training Details
 
-**BibTeX:**
+### Pre-training
 
-[More Information Needed]
+The model uses the pretrained [all-distilroberta-v1](https://huggingface.co/sentence-transformers/all-distilroberta-v1).
 
-**APA:**
+### Fine-tuning
 
-[More Information Needed]
+The model is fine-tuned for multi-class relevance classification based on the ESCI framework. The fine-tuning process involves an input of query-product pairs, and an objective of classification using cross entropy loss to align predicted class probabilities with true labels.
 
-## Glossary [optional]
+### Hyperparameters
 
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+Training was performed on a Tesla V100-PCIE-32GB GPU with a batch size of 32 over 3 epochs. The learning rate was set to 5e-5 and optimized using the AdamW optimizer, with 10% of the total training steps allocated for warm-up. Input sequences were padded to a max length of 512 tokens. Validation was conducted every ~10% of an epoch, and micro F1 score and accuracy were used to evaluate performance.
 
-[More Information Needed]
+### Training Data
 
-## More Information [optional]
+<!-- 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]
+| Dataset                                                  | Paper                                    | Number of training tuples  |
+|--------------------------------------------------------|:----------------------------------------:|:--------------------------:|
+| [Amazon Shopping Queries Dataset](https://github.com/amazon-science/esci-data/tree/main/shopping_queries_dataset) | [paper](https://arxiv.org/pdf/2206.06588) | 1,253,756 |
 
-## Model Card Authors [optional]
+## Model Card Authors
 
-[More Information Needed]
+[Sarah Lawlis](https://www.linkedin.com/in/sarah-lawlis/)
 
 ## Model Card Contact
 
-[More Information Needed]
+[email protected]