fine-tuned-distilbert-base-uncased-LLM-Judge

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->

Model Details

Model Description

<!-- 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.

• Developed by: [Ruiqi Tan]
• Model type: [Text categorization]
• License: [Apache 2.0]
• **Finetuned from model [distilbert-base-uncased]

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

The task is to predict user preferences based on provided input prompts and chatbot responses.

Out-of-Scope Use

<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->

[To estimate wihch text could be more prefereable based on large scale survey]

Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

[Language and usage cultural zone and education level bias]

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. -->

[The training data can be found on Kaggle competition page: https://www.kaggle.com/competitions/llm-classification-finetuning/data]

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]

[The dataset is processed to encode model_a and model_b as embeddings, while tokenizing textual data using DistilBERT's tokenizer.]

Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

The dataset was split into 80% training and 20% testing. The categorical embeddings were trained alongside the transformer model for joint optimization.

Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

[At the end of each epoch, validation loss and accuracy were calculated to evaluate the model's performance. These metrics are included in the repository.]

Results

[The model achieved accuracy arund 45%]

Model Architecture and Objective

[The solution combines the power of a pretrained transformer model with categorical embeddings:

Textual Features: The prompt, response_a, and response_b were tokenized using the distilbert-base-uncased tokenizer and processed through a pretrained DistilBERT model.

Categorical Features: The categorical variables model_a and model_b were encoded into numeric IDs and processed using separate embedding layers.

Combined Features: The outputs from the transformer model (textual features) and the embeddings (categorical features) were concatenated and passed through a fully connected classifier to predict user preference.]

Hardware

[Hardware Overview:

  Model Name: MacBook Pro
  Model Identifier: MacBookPro16,2
  Processor Name: Quad-Core Intel Core i7
  Processor Speed: 2,3 GHz
  Number of Processors: 1
  Total Number of Cores: 4
  L2 Cache (per Core): 512 KB
  L3 Cache: 8 MB
  Hyper-Threading Technology: Enabled
  Memory: 32 GB
  System Firmware Version: 2069.80.3.0.0 (iBridge: 22.16.13040.5.1,0)
  OS Loader Version: 582~3227
  Serial Number (system): C02CX487ML85
  Hardware UUID: 23703B16-1430-55A6-96D3-B1D3121CDD33
  Provisioning UDID: 23703B16-1430-55A6-96D3-B1D3121CDD33
  Activation Lock Status: Enabled]

Model Card Contact

[More Information Needed]