* modeld_v2: spatial features
* Update fetcher.py
* dont reshape non 4 dim arrays
* realize for non compiled
* Update compile_modeld.py
* god dammit it was realize()
* it was fucking frozen tinygrad. just need to recompile
* bump
* ci: add is_big flag to metadata.json to support backward compat
* Update model_generator.py
* Update sunnypilot-build-model.yaml
* Update helpers.py
* Revert "Update helpers.py"
This reverts commit 3a955ca11a.
* Reapply "Update helpers.py"
This reverts commit ca9c6e1933.
* models: use less strict chestnut detection state
---------
Co-authored-by: Jason Wen <haibin.wen3@gmail.com>
* ci: Dedicated Model Runner
* recurse
* not needed
* fix wrapper
* whoops
* bypass
* modeld_v2: restore chestnut link check before big model build
* modeld_v2: stage onnx to disk instead of shared memory
* ci: clear unchunked onnx temps before model build
* ci: stream the pkl hash instead of loading it into memory
---------
Co-authored-by: Jason Wen <haibin.wen3@gmail.com>
* ci: fix prebuilt file copy for null-separated release_files.py output
* ci: drop prebuilt symlinks the launch script recreates and gate the release on no submodules
* ci: use the device-local scons cache and prune dead prebuilt config
* ci: keep the scons cache in the runner workspace instead of the device's
* modeld_v2: Support eGpu
* bump tg
* egpu
* no pkls please
* god no onnx either
* fix test
* done in build model now
* lint
* rip
* egpu ready build all split
* manual seed , reuse memory buffers across runs
* dont download big when we dont have big lol
* whoops
* no i and x
* cd
* who put those there. ??
* reduce flakiness by using artifact-name from build-model to regex, speed up pub b y checking the name before trying to clone and publish again
* try hf as a trusted publisher :)
* mf its a dataaset. i knew that
* fucking validation wants raw to fetch and full to push. grr
* smh
* dude i am missing so much
* pkl name
* move build all to hf
* tests: migrate sunnypilot tests to unittest and remove pytest
* red diff mf
* im scared , this may be a bad idea lol
* fetch latest commit.
* transition to requests
* models: use requests instead of aiohttp
* tici
* fix
* gpu fixes from upstream
* lint
* bump
* needed to say
* old
* how??
* epgu flag reduce
* this made me cry
* support monolith still
* precache warp in legacy
* Move jsons to param for sunnylink
---------
Co-authored-by: Jason Wen <haibin.wen3@gmail.com>
* modeld_v2: safe model validation
* fix string
* numpy
* dumb
* god use full attribute names please
* modeld_v2: refactor compile_modeld
* redundant
* CREAM AND SUGAR
* gpu stuffs
* Update fetcher.py
* Update compile_modeld.py
* Update compile_modeld.py
* Update compile_modeld.py
* Update compile_modeld.py
* summary
* Update compile_modeld.py
* Update compile_modeld.py
* Update compile_modeld.py
* i could lie say
* needed
* i could
* Update compile_modeld.py
* done done done
* bye metadata
* simplify
* deeeeep
* oopsie
* i don't know what i'm doing
* it's a supercombo
* wtf. ghostwriter
* i might be blind
* fuck. i AM blind. or dumb. or both.
* hmmmm
* read/open - what's the difference
* realize. that i don't know shit
* whatever
* fix paths
* fuckit. dynamic everything.
* simplify everything.
* fix macos cabana
* np.random.seed is deprecated, use Generator or default rng instead @nayan
* this is annoying me
* too much fluff
* no
* lint be crazy now, man i've been away a while
* back
* fix pkl loader test
* that comment was wrong, its still per model, just compiled at with the input/output shapes
* fix chunking
* dump and assign to cpu
* prev_feat in cpu to prevent corruption
* add todo-sp
* desire me?
* allow legacy
* lint
* shapey
* bye
---------
Co-authored-by: discountchubbs <alexgrant990@gmail.com>
Co-authored-by: James Vecellio-Grant <159560811+Discountchubbs@users.noreply.github.com>
Co-authored-by: Jason Wen <haibin.wen3@gmail.com>
* zero ll patched big model
* probe in a subprocess so usbgpu lock gets released
* compiles
* runs
* num_jobs gets overwritten, use side effect
* poll tg devices
* make sure build crashes on missing gpu
* fine not to rely on Device.default
* seperate tg env for each model runner
* comment
* Revert "seperate tg env for each model runner"
This reverts commit f6470cc4258eaeb3e8e37907ef370871c9af5aa4.
* env is shared, gate on flag
* no fallback warp dev must be set
* build for current device only, unless pc/release
* comment
* list
* listen for plug in
* add icon to status bar, read params on every frame (?)
* log available devices
* try copy out when loading?
* Revert "log available devices"
This reverts commit e8c52a5d59456d4820ecb13b99a6c46ea1386a20.
* Revert "try copy out when loading?"
This reverts commit 518f403aa03faeda1950fe3dbce0d9e4c1584455.
* don't trigger device probe/caching on modeld prepare
* re-export with ll and road edges
* dont cache devices in manager process
* get USBGPU from params
* no usbgpu env
* missed one
* sconscript don't poll
* unconditional env
* always explicitely set devices on input tensors
* set DEV so amd uses right compiler and iface??
* fix flag
* bump tg
* rm xdg_cache_home
* tg don't bump all the way
* missing gmmu=0 at compile time
* dm set dev
* tg backend
* update gitignore
* missing import
* unused imports
* rely on Device.DEFAULT at compile time (already the case bc onnxrunner)
* comments
* dm warp needs DEV set too
* build both smol and big
* misc typos
* set dev at compile time
* don't need
* DEV=CPU when getting metadata, ensure we don't grab gpu lock
* this would also grab lock
* put bool
* warp compile always prepare only
* missed one
* poll ui
* missing here
* don't force usbgpu at build time
* tmp patch fetch_fw
* catch all, follow hardwared patterns
* simpler
* compile make input queues
* revert this
* group this more readable
* rm empty line
* make dummy frame using numpy
* revert compile make input queues
* no compiler at runtime
* cleanup
* fine to rebuild all on change to device node for now
* fix usbgpu_present
* fix sconscript
* no size in header stream decompress
* DEBUG=2
* minimal viable feedback
* egpu gray
* oops
* gotta do this actually
* modeld build only depends on modeld devices
* don't ship onnx to release? or chunk
* don't need
* can only set compiler on dev=
* none device works, will use default
* make linter happy
* chunk agnostic onnx input to compile_modeld
* chunk big onnx
* +x chunker
* fix #!
* and don't ship chunked onnx to release
* firmware now in correct location
* better err on missing onnx/chunk
* SConscript also need to accept chunked onnx
* metadata also need to load maybe chunked
* dedupe cmd
* this needs to be on cpu
* devices are set in the tgflags, we already depend on them
* rebuilding on changed order is fine
* read file chunked can already load either chunked or not
* chunk all big onnx
* less confusing
* unused import
* python device to load onnx bytes
* default device for runners, python for metadata
* why not
* chunked to shm
* version: a new beginning
* changelog
* singular
* show ours
* actual
* readjust
* updated
* more
* official spelling
* more
* sync
* fix
* send it
* push
* we never had this lol
* syncs
* ci: add validate-test-on-staging-c3 branch to deployment triggers and enhance stable branch handling
* fix long overdue mistake lol
* ci: add condition to wait for start on push events in build workflow
* Fix extra version identifier
* no need for this, i validated what I needed
* only care for release tags, not any
* fix: update versioning logic to use build date and run number for tagging
* fix: update tagging logic and enhance commit message format in build scripts
* fix: refine tagging condition to exclude tag pushes for stable branches
* fix: add extra version identifier to output for better version tracking
* trying to keep things clean and simple
* bugfix
James Vecellio-Grant