- Fix clustering strange behavior
- Clustering inconsistencies, errors
- strange clustering errors
- Extensive documentation
- Code is cryptic, esoteric, and generally confusing in both appearance and function.
- Draft a getting started page.
- Finalize live node system
- Tying changing information (like the current time) to static nodes for quick reference by the semweb
- Data flows
- Data must flow upwards from textflow for communication between loops
- Connect the language loop to the causal model
- Reconnect the speaker-id to textflow and update the semantic web to accept it again
- Forgetfulness
- Semantic web needs to forget conflicting facts to improve accuracy and time efficiency
- Dependent on completion of root representation.
- Runtime
- Call each loop from a main one.
- Deployment
- Update github
- Located and removed bug in clusterResolve
- Improved Wyze integrations
- Began root representation format
- Optimized audio loop
- Fixed up the repo / serious misplaced code issue
- Organized a little bit
- More persistency
- Fix clustering strange behavior
- Clustering inconsistencies, errors
- strange clustering errors
- Extensive documentation
- Code is cryptic, esoteric, and generally confusing in both appearance and function.
- Draft a getting started page.
- Finalize live node system
- Tying changing information (like the current time) to static nodes for quick reference by the semweb
- Data flows
- Data must flow upwards from textflow for communication between loops
- Connect the language loop to the causal model
- Reconnect the speaker-id to textflow and update the semantic web to accept it again
- Forgetfulness
- Semantic web needs to forget conflicting facts to improve accuracy and time efficiency
- Dependent on completion of root representation.
- Runtime
- Call each loop from a main one.
- Deployment
- Update github
- Rewrite of clusterResolve
- Began Electron app
- Runtime working
- Fixed up the repo / serious misplaced code issue
- Tested language processing at scale
- Fuzzy date, time, extraction, resolution, profile resolution, reference for both, symbolic pre-processing, symbolic pre-pre-processing
- Inflection standardization
- Discovered necessity of editing, rerolling wern, editing, rerolling wern,...
- part 1/3 of nprop implementation (where in runtime, preprocessing)
- Reliable fast TTS
- Setup: pip install pipwin && pipwin install pyaudio && python cleo.py -s "text to be spoken"
- Install, setup, script
- Failed dockerization (too much change)
- Resetting repo again (too much change)
- shelving UI until post github
- symbolically parsing time is probably NP-hard
- so close to getting to production level
- QA accuracy, recall accuracy, increases as # of symbolic conversions increase
- fixing up repo is halfway done (dead-end imports gone)
- symbolically parsing time is still probably NP-hard
- substitution reliant on balancing edits versus accuracy, make it make sense
- "dockerized" working partially, unforeseen drawbacks
- symbolic standardization, states talking to calculus, hugely reliant standardization
- charting things
- added punctation handling for playing :, ,, !, ?, .
- fixed bug with playing start of audio experiencing clipping in non-blocking hardware by adding a slight delay
- cleaning up the functionality of adding pauses before/after/between words
- started using the pyaudio interface directly to have our own functions for playing back the generated audio segments
- pulls
- some concluding thoughts:
- model architecture json is necessary, it is a language model that gets generated in the wrong place but I don't want to fix it right now
- Kaldi changed versions and changed the recognizer call
- svomapping.pdf
- made landing more aesthetically pleasing
- made a readthedocs
- demo
- amount of integrations causing headaches
- large scope of use, memory; accuracy breaking down
- Added flags.json to track initialization state as well as enabled integrations
- Added Dockerfile
- Added dockerize.sh to automatically build docker image
- Docker image builds to ~ 6.5 gb, using python3 base
- Known Tensorflow version bug (Thanks to PIP)
- Furthered SVO mappings
- Identified multiple conflicts between dev repo and this one
- Advanced nprop formal system design significantly (experimentally implementing)
- Began to expedite workflow based around dev repo -> this one -> automated docker build