Thomas Capelle

Diffusion-Models-pytorch. Pytorch implementation of Diffusion Models (https://arxiv.org/pdf/2006.11239.pdf)

timeseries_fastai. fastai V2 implementation of Timeseries classification papers.

apple_m1_pro_python. A collection of ML scripts to test the M1 Pro MacBook Pro

llm_recipes. A set of scripts and notebooks on LLM finetunning and dataset creation

termGPT. A simple wrapper to call openAI's chatGPT on the terminal written in Python

cloud_diffusion. Diffusion on the Clouds: Short-term solar energy forecasting with Diffusion Models

moving_mnist. Exploring Moving Mnist dataset with forecasting algorithms

metnet_pytorch. A MetNet implementation in Pytorch and fastai

action_recognition. Solving UCF-101 with fastai2

gpt_translate. Translate a docodile website using GPT4

fastai_timm. Exploring fastai and timm integration for model finetune

mixtral. Mixtral finetuning

whisper. Exporting youtube videos using whisper

torch_moving_mnist. A simple Dataset generator for Moving Mnist

eclipse_pytorch. My take on ECLIPSE solar nowcasting DL paper

mistral_wandb. A full fledged mistral+wandb

simple_agent. A simple GPT-powered agent to help you research

solar_seg. Solar segmentation of PV modules

torchdata. Exploring new PyTorch 1.11 data API

minai. A miniture AI training framework for PyTorch

resnet_checkpoints. Exploring gradient chekcpoints resnets with fastai V2

pokepalette. A color library based on pokemons colors!

sagemaker-llama2-wandb. Jupyter Notebook

shear. Shear and Pruning of LLMs

hydra_net. A simple HydraNet for depth and segmentation

bench_llms. A simple benchmark of LLM speed

mini_llm. A minimal training/eval LLM repo

bedrock_weave. An example on using wandb.weave with Bedrock models

cloud_segmentation. A simple segmentation model on public images, with wandb support

hackercup_rag. A simple RAG pipeline to get you started

dummy_docs. A repo with some docs

nanollada. A simple nanoGPT 2 Llada

rocket. This is a multi-channel implementation of ROCKET

mini_server. A simple fastapi routing server to expose a model endpoint

ETHz. A repo for the tools needed for the Hackathon

grpo_weave. A starter code to GRPO with great telemetry

PhyDNet. Code for our CVPR 2020 paper "Disentangling Physical Dynamics from Unknown Factors for UnsupervisedVideo Prediction"

alignment-handbook. Robust recipes for to align language models with human and AI preferences

tcapelle.github.io. my website

TGS. TGS kagle challenge

wandbot_llm. A OSS LLM trained on wandb API

fmnist. A benchmark on Fashion MNIST

KernelBench. KernelBench: Can LLMs Write GPU Kernels? - Benchmark with Torch -> CUDA problems

cw-axolotl. Simple k8s launcher

amazon-bedrock-agent-samples. Example Jupyter notebooks 📓 and code scripts 💻 for using Amazon Bedrock Agents 🤖 and its functionalities

eval_vlm. A quick eval of recent Llama 3.2

claudio. A claude code clone