mchughes288 Goto Github PK

babyagi

berkeley_deep_rl

Assignments for Berkeley CS 285: Deep Reinforcement Learning (Fall 2020)

chat_booth

Speech-to-text, ChatGPT and text-to-speech.

conf_scores

Map confidence scores

dotfiles

Easily deploy my zsh and tmux configuration on new machines. Includes local and remote aliases to improve workflow.

emotion_detection_cpc

Emotion detection in audio utilising self-supervised representations trained with Contrastive Predictive Coding (CPC).

evals_template

Template for any evals project using LLM apis

harmbench

HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

hqa_jukebox

OpenAI style jukebox using HQA to encode audio into discrete codes to do sequence modelling over.

idp

Integrated Design Project: Develop an AGV to collect, identify and deliver pallets to drop off zones in an enclosed arena. Claimed first prize in this 2nd year Engineering competition at Cambridge.

meaning_error_rate

Meaning Error Rate (MER) is an alternative metric for evaluating speech recognition systems that considers changes in meaning. It is automated by using GPT-3, few shot learning, and chain of thought.

model_written_evals

Create lm evals to look for inverse scaling behaviour

quote_button

Raspberry pi powered quote button that plays sound bites in response to physical, visual and audio inputs.

rt_speech_summariser

Hook up Speechmatics RT transcription to GPT3 API.

sensortag-fpv

App that analyses FPV drone freestyle with the TI SensorTag CC2650. Final project for the 4B25 engineering module at Cambridge University.

speechmatics-python

Python library and CLI for Speechmatics

sunflower-simulator

Sunflower Full-System Hardware Emulator and Physical System Simulator for Sensor-Driven Systems.

text2art

This project aimed to compile and test various text-to-art resources using the VQGAN and DALLE generative models. The output of these models can be improved with ESRGAN upscaling.

torchpq

Efficient implementations of Product Quantization and its variants using Pytorch and CUDA

voice_cloning

Clone and transcribe your voice in 5 seconds and generate a synthesised continuation with GPT-2. Many new features compared to forked repo.

warp-firmware

Firmware for the Cambridge Physical Computation Laboratory's Warp Embedded Multi-Sensor Platform.

whisper_logit_lens

This Alignment Jam Hackathon project explores whether the concept of "logit lens" applies to the encoder and decoder layers in Whisper, an end-to-end speech recognition model.