matpalm Goto Github PK

1e6_sentences

simple parsing / embedding utils for the 1e6 sentences dataset

aliens_drum_kit

building sample embeddings so we can extract a drum kit from the audio of "aliens"

arduino_hacking

audioscrobbler-experiments

finding the shortest path from napalm death to celion dion through audioscrobbler associations

bipartite-dispersion

bnn

bee detection tensorflow conv net for a rasp pi on side of a hive

bullet_grasping

hacking with the bullet grasping setup

burnit

the dvd packing problem

cached_dilated_causal_convolutions

1D dilated causal convolutions with extreme caching for fast inference on a daisy patch ( cortex m7 ) or eurorack pmod ( FPGA )

cartpoleplusplus

3d cartpole gym env using bullet physics trained from pixels with tensorflow LRPG, DDPG & NAF

collocations

bigram / trigram analysis of wikipedia; mainly mutual info

common-crawl

playing around with the common crawl dataset

common-crawl-quick-hacks

common crawl quick hack examples

connected-components

test of iterative python embedded pig

consistent_hash

afternoon hackery visualising the consistent hash method

daisyexamples

Examples for the Daisy Platform

dead-simple-ga

a dead simple genetic algorithm for whietpages book club

dets_test

testing scalability aspects of erlang dets tables

dither_net

a GAN experiment to generate dithers minimising pixel change between frames for a very slow movie player

diy_twitter_client

a simple twitter client that learns what i like to read

dm_pix

PIX is an image processing library in JAX, for JAX.

doing_bayesian_data_analysis

drivebot

tensorflow deep RL for driving a rover around

embedding-cache

local file cache over tf.hub resnet inception feature vectors

embedding_the_chickens

ensemble_net

training ensembles with jax

eurorack-pmod

Hardware and gateware for getting started in FPGA-based audio synthesis with open source tools.

evolved_cartpole

evolving a gym cartpole controller; directly against bytes of a tf lite microcontroller model

evolved_channel_selection

rather than use all 13 channels in a multi spectral image for classification can we train a model that is robust to all combos, at all resolutions, and use a genetic algorithm to choose which are the most valuable?

farmbot_api

python api for bulk farm bed sweeps, image capture and plant detection