ShahzadSohail
/

emotion_detection_model_final

@@ -1,210 +1,140 @@
 ---
 language: en
 tags:
-  - text-classification
-  - pytorch
-  - roberta
-  - emotions
-  - multi-class-classification
-  - multi-label-classification
 datasets:
-  - go_emotions
 license: mit
 widget:
-  - text: I am not having a great day.
 ---
-# 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:** [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]
-<!-- Provide the basic links for the model. -->
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
-## 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]
-## 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]

 ---
 language: en
 tags:
+- text-classification
+- pytorch
+- roberta
+- emotions
+- multi-class-classification
+- multi-label-classification
 datasets:
+- go_emotions
 license: mit
 widget:
+- text: I am not having a great day.
 ---
+#### Overview
+Model trained from [roberta-base](https://huggingface.co/roberta-base) on the [go_emotions](https://huggingface.co/datasets/go_emotions) dataset for multi-label classification.
+#### How the model was created
+The model was trained using `AutoModelForSequenceClassification.from_pretrained` with `problem_type="multi_label_classification"` for 3 epochs with a learning rate of 2e-5 and weight decay of 0.01.
+#### Inference
+There are multiple ways to use this model in Huggingface Transformers. Possibly the simplest is using a pipeline:
+```python
+from transformers import pipeline
+classifier = pipeline(task="text-classification", model="ShahzadSohail/emotion_detection_model_final", top_k=None)
+sentences = ["I am not having a great day"]
+model_outputs = classifier(sentences)
+print(model_outputs[0])
+# produces a list of dicts for each of the labels
+```
+##### Summary
+As provided in the above notebook, evaluation of the multi-label output (of the 28 dim output via a threshold of 0.5 to binarize each) using the dataset test split gives:
+- Accuracy: 0.474
+- Precision: 0.575
+- Recall: 0.396
+- F1: 0.450
+But the metrics are more meaningful when measured per label given the multi-label nature (each label is effectively an independent binary classification) and the fact that there is drastically different representations of the labels in the dataset.
+With a threshold of 0.5 applied to binarize the model outputs, as per the above notebook, the metrics per label are:
+|                | accuracy | precision | recall | f1    | mcc   | support | threshold |
+| -------------- | -------- | --------- | ------ | ----- | ----- | ------- | --------- |
+| admiration     | 0.946    | 0.725     | 0.675  | 0.699 | 0.670 | 504     | 0.5       |
+| amusement      | 0.982    | 0.790     | 0.871  | 0.829 | 0.821 | 264     | 0.5       |
+| anger          | 0.970    | 0.652     | 0.379  | 0.479 | 0.483 | 198     | 0.5       |
+| annoyance      | 0.940    | 0.472     | 0.159  | 0.238 | 0.250 | 320     | 0.5       |
+| approval       | 0.942    | 0.609     | 0.302  | 0.404 | 0.403 | 351     | 0.5       |
+| caring         | 0.973    | 0.448     | 0.319  | 0.372 | 0.364 | 135     | 0.5       |
+| confusion      | 0.972    | 0.500     | 0.431  | 0.463 | 0.450 | 153     | 0.5       |
+| curiosity      | 0.950    | 0.537     | 0.356  | 0.428 | 0.412 | 284     | 0.5       |
+| desire         | 0.987    | 0.630     | 0.410  | 0.496 | 0.502 | 83      | 0.5       |
+| disappointment | 0.974    | 0.625     | 0.199  | 0.302 | 0.343 | 151     | 0.5       |
+| disapproval    | 0.950    | 0.494     | 0.307  | 0.379 | 0.365 | 267     | 0.5       |
+| disgust        | 0.982    | 0.707     | 0.333  | 0.453 | 0.478 | 123     | 0.5       |
+| embarrassment  | 0.994    | 0.750     | 0.243  | 0.367 | 0.425 | 37      | 0.5       |
+| excitement     | 0.983    | 0.603     | 0.340  | 0.435 | 0.445 | 103     | 0.5       |
+| fear           | 0.992    | 0.758     | 0.603  | 0.671 | 0.672 | 78      | 0.5       |
+| gratitude      | 0.990    | 0.960     | 0.881  | 0.919 | 0.914 | 352     | 0.5       |
+| grief          | 0.999    | 0.000     | 0.000  | 0.000 | 0.000 | 6       | 0.5       |
+| joy            | 0.978    | 0.647     | 0.559  | 0.600 | 0.590 | 161     | 0.5       |
+| love           | 0.982    | 0.773     | 0.832  | 0.802 | 0.793 | 238     | 0.5       |
+| nervousness    | 0.996    | 0.600     | 0.130  | 0.214 | 0.278 | 23      | 0.5       |
+| optimism       | 0.972    | 0.667     | 0.376  | 0.481 | 0.488 | 186     | 0.5       |
+| pride          | 0.997    | 0.000     | 0.000  | 0.000 | 0.000 | 16      | 0.5       |
+| realization    | 0.974    | 0.541     | 0.138  | 0.220 | 0.264 | 145     | 0.5       |
+| relief         | 0.998    | 0.000     | 0.000  | 0.000 | 0.000 | 11      | 0.5       |
+| remorse        | 0.991    | 0.553     | 0.750  | 0.636 | 0.640 | 56      | 0.5       |
+| sadness        | 0.977    | 0.621     | 0.494  | 0.550 | 0.542 | 156     | 0.5       |
+| surprise       | 0.981    | 0.750     | 0.404  | 0.525 | 0.542 | 141     | 0.5       |
+| neutral        | 0.782    | 0.694     | 0.604  | 0.646 | 0.492 | 1787    | 0.5       |
+Optimizing the threshold per label for the one that gives the optimum F1 metrics gives slightly better metrics - sacrificing some precision for a greater gain in recall, hence to the benefit of F1 (how this was done is shown in the above notebook):
+|                | accuracy | precision | recall | f1    | mcc   | support | threshold |
+| -------------- | -------- | --------- | ------ | ----- | ----- | ------- | --------- |
+| admiration     | 0.940    | 0.651     | 0.776  | 0.708 | 0.678 | 504     | 0.25      |
+| amusement      | 0.982    | 0.781     | 0.890  | 0.832 | 0.825 | 264     | 0.45      |
+| anger          | 0.959    | 0.454     | 0.601  | 0.517 | 0.502 | 198     | 0.15      |
+| annoyance      | 0.864    | 0.243     | 0.619  | 0.349 | 0.328 | 320     | 0.10      |
+| approval       | 0.926    | 0.432     | 0.442  | 0.437 | 0.397 | 351     | 0.30      |
+| caring         | 0.972    | 0.426     | 0.385  | 0.405 | 0.391 | 135     | 0.40      |
+| confusion      | 0.974    | 0.548     | 0.412  | 0.470 | 0.462 | 153     | 0.55      |
+| curiosity      | 0.943    | 0.473     | 0.711  | 0.568 | 0.552 | 284     | 0.25      |
+| desire         | 0.985    | 0.518     | 0.530  | 0.524 | 0.516 | 83      | 0.25      |
+| disappointment | 0.974    | 0.562     | 0.298  | 0.390 | 0.398 | 151     | 0.40      |
+| disapproval    | 0.941    | 0.414     | 0.468  | 0.439 | 0.409 | 267     | 0.30      |
+| disgust        | 0.978    | 0.523     | 0.463  | 0.491 | 0.481 | 123     | 0.20      |
+| embarrassment  | 0.994    | 0.567     | 0.459  | 0.507 | 0.507 | 37      | 0.10      |
+| excitement     | 0.981    | 0.500     | 0.417  | 0.455 | 0.447 | 103     | 0.35      |
+| fear           | 0.991    | 0.712     | 0.667  | 0.689 | 0.685 | 78      | 0.40      |
+| gratitude      | 0.990    | 0.957     | 0.889  | 0.922 | 0.917 | 352     | 0.45      |
+| grief          | 0.999    | 0.333     | 0.333  | 0.333 | 0.333 | 6       | 0.05      |
+| joy            | 0.978    | 0.623     | 0.646  | 0.634 | 0.623 | 161     | 0.40      |
+| love           | 0.982    | 0.740     | 0.899  | 0.812 | 0.807 | 238     | 0.25      |
+| nervousness    | 0.996    | 0.571     | 0.348  | 0.432 | 0.444 | 23      | 0.25      |
+| optimism       | 0.971    | 0.580     | 0.565  | 0.572 | 0.557 | 186     | 0.20      |
+| pride          | 0.998    | 0.875     | 0.438  | 0.583 | 0.618 | 16      | 0.10      |
+| realization    | 0.961    | 0.270     | 0.262  | 0.266 | 0.246 | 145     | 0.15      |
+| relief         | 0.992    | 0.152     | 0.636  | 0.246 | 0.309 | 11      | 0.05      |
+| remorse        | 0.991    | 0.541     | 0.946  | 0.688 | 0.712 | 56      | 0.10      |
+| sadness        | 0.977    | 0.599     | 0.583  | 0.591 | 0.579 | 156     | 0.40      |
+| surprise       | 0.977    | 0.543     | 0.674  | 0.601 | 0.593 | 141     | 0.15      |
+| neutral        | 0.758    | 0.598     | 0.810  | 0.688 | 0.513 | 1787    | 0.25      |
+This improves the overall metrics:
+- Precision: 0.542
+- Recall: 0.577
+- F1: 0.541
+Or if calculated weighted by the relative size of the support of each label:
+- Precision: 0.572
+- Recall: 0.677
+- F1: 0.611
+#### Commentary on the dataset
+Some labels (E.g. gratitude) when considered independently perform very strongly with F1 exceeding 0.9, whilst others (E.g. relief) perform very poorly.
+This is a challenging dataset. Labels such as relief do have much fewer examples in the training data (less than 100 out of the 40k+, and only 11 in the test split).
+But there is also some ambiguity and/or labelling errors visible in the training data of go_emotions that is suspected to constrain the performance. Data cleaning on the dataset to reduce some of the mistakes, ambiguity, conflicts and duplication in the labelling would produce a higher performing model.