mnist_nnn_vision / README.md

Update README.md

5c06078 about 1 year ago

6.12 kB

	---
	license: apache-2.0
	datasets:
	- mnist
	metrics:
	- accuracy
	pipeline_tag: image-classification

	model-index:
	- name: mnist_nnn_vision
	results:
	- task:
	type: image-classification # Required. Example: automatic-speech-recognition
	name: Image Classification # Optional. Example: Speech Recognition
	dataset:
	type: mnist # Required. Example: common_voice. Use dataset id from https://hf.co/datasets
	name: MNIST # Required. A pretty name for the dataset. Example: Common Voice (French)
	split: test # Optional. Example: test
	metrics:
	- type: accuracy # Required. Example: wer. Use metric id from https://hf.co/metrics
	value: 0.9311 # Required. Example: 20.90
	name: Accuracy # Optional. Example: Test WER
	verified: true # Optional. If true, indicates that evaluation was generated by Hugging Face (vs. self-reported).
	---

	# Model Card for NNN (Not a Neural Network)

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

	This modelcard aims to be a base template for new models. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/modelcard_template.md?plain=1).
	Just a simple exercise I did to learn how to use the PyTorch and TorchHD libraries
	## Model Details

	### Model Description

	<!-- Provide a longer summary of what this model is. -->

	This MNIST model was made using 2 libraries: PyTorch and TorchHD.
	The HD in TorchHD stands for Hyperdimensional Computing, which means TorchHD is a library that allows you to do hyperdimensional computing in PyTorch.
	Hyperdimensional Computing (Or HDC) models are much less accurate than neural networks, that's why this model's accuracy is ~82%

	- Developed by: Comrade Cat (me)
	- Shared by: Comrade Cat (me)
	- Model type: Image Classification
	- Language(s) (NLP): None
	- License: Apache 2.0
	- Finetuned from model: None. This is a pretrained model.

	### Model Sources [optional]

	<!-- Provide the basic links for the model. -->

	- Repository: Here
	- Paper: None
	- Demo: Not available yet.

	## 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. -->
	This model is intended to be used as an experiment to compare TorchHD models to PyTorch models.

	### Direct Use

	<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
	This model is intended to be used for recognizing digits. Please be aware that it has a lower accuracy than a normal PyTorch model.

	### Downstream Use

	<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->

	This model could be fine-tuned to improve its accuracy, as it is surprisingly low.

	### Out-of-Scope Use

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

	Please do not misuse the model. This model will not work for tasks other than handwritten digit recognition.

	## Bias, Risks, and Limitations

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

	This model is too simple and inaccurate to be biased against a social group.
	The technical limitations are its inaccuracy.

	### Recommendations

	<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

	Users (both direct and downstream) should be aware of the risks, biases and limitations of the model.
	Be aware of how inaccurate this model is!!!

	## How to Get Started with the Model

	Download both the model and the encoder. Make sure to download their weights too if you want to fine-tune them!
	After that you can load them in PyTorch.

	```python
	import torch

	# Load the base model and weights
	model = torch.load("mnist.pt")
	model.load_state_dict(torch.load("mnist_weights.pt"))

	# Load the encoder and its weights
	encoder = torch.load("mnist_encoder.pt")
	encoder.load_state_dict("mnist_encoder_weights.pt")

	# Load an image of a handwritten digit.
	# sample_image = (load your image here)

	# Encode the loaded image
	encoded_image = encode(sample_image)
	outputs = model(encoded_image)
	print(outputs)
	```

	## Training Details

	### Training Data

	<!-- This should link to a Data 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. -->

	[Link to MNIST will be added soon]

	### 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

	[More Information Needed]


	#### Training Hyperparameters

	- Training regime: [I don't know yet] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
	- DIMENSIONS: 11000
	- IMAGE SIZE: 28
	- NUMBER OF LEVELS: = 1000
	- BATCH SIZE: 2
	#### Speeds, Sizes, Times

	<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->

	The training of this model took 1 hour, because I have a potato PC

	## Evaluation

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

	### Testing Data, Factors & Metrics

	#### Testing Data

	<!-- This should link to a Data Card if possible. -->

	[Link to MNIST will be added soon]

	#### 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. -->

	[Accuracy: 82.850%]

	### Results

	[Low accuracy]

	#### Summary

	This model is simply too inaccurate for its own good. However, I (Comrade Cat), will try to retrain the model until it has better accuracy.

	## Model Card Contact

	[More Information Needed]