Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
83 commits
Select commit Hold shift + click to select a range
992c77c
refactor(llm): remove native C++ LLM runner and tests
barhanc Oct 2, 2026
e8a47f6
refactor(bridge): remove llm namespace from rnexecutorchJsi docstring
barhanc Oct 2, 2026
9fe43d3
style
barhanc Oct 2, 2026
2ae8fa4
feat(model): support returning primitive lists from execute
barhanc Oct 2, 2026
ec532a6
refactor(llm): split runner into runners/ directory and update stats …
barhanc Oct 2, 2026
87f4439
feat(llm): add runners/ directory with textRunner, multimodalRunner, …
barhanc Oct 2, 2026
731edfd
style
barhanc Oct 2, 2026
4ec801d
style
barhanc Oct 2, 2026
7961f06
feat(llm): add sampling pipeline (wip)
barhanc Oct 2, 2026
d9eaa1e
feat(llm): add token sampling pipeline and wire into textRunner
barhanc Oct 3, 2026
6d4b0ed
wip
barhanc Oct 3, 2026
3b87789
revert(model): revert primitive list support from execute and bridge …
barhanc Oct 3, 2026
3d27e96
style
barhanc Oct 3, 2026
16d3d16
feat(llm): add native C++ sampler and install map
barhanc Oct 5, 2026
fecf0fd
style(llm): drop churn from sampler wiring
barhanc Oct 5, 2026
d2563c6
fix(tests): drop et_llm_runner and LlmRunnerTest from the test build
barhanc Oct 5, 2026
afc7ded
refactor(llm): prefix runner modules and drop math multinomial
barhanc Oct 5, 2026
a69ea58
refactor(llm): clean up default sampling options formatting
barhanc Oct 6, 2026
0182052
docs(llm): emphasize in-place logits modification in sample JSDoc
barhanc Oct 6, 2026
e868dfa
refactor(llm): remove profiling code from llmTextRunner
barhanc Oct 6, 2026
5a040bb
style
barhanc Oct 7, 2026
5fdec2d
feat(llm): implement multimodal runner and clean up text runner
barhanc Oct 7, 2026
07695a3
fix(llm): allow rank-4 targetShape in chatPreprocessor
barhanc Oct 7, 2026
5368533
refactor(llm): decouple specialized runner specs and polymorphically …
barhanc Oct 7, 2026
ab82eb1
fix(llm): add descriptive error message for unsupported media segment…
barhanc Oct 7, 2026
e7cdb69
refactor(llm): simplify segment kind access in prefill error
barhanc Oct 7, 2026
2c63dd0
refactor(llm): flatten specialized runner specs into model specs
barhanc Oct 7, 2026
fc0ba83
refactor(llm): self-validate runner specs inside specialized runner c…
barhanc Oct 7, 2026
964eb57
style
barhanc Oct 7, 2026
2ce17cc
style
barhanc Oct 7, 2026
f13dc53
style
barhanc Oct 7, 2026
7c74aa9
update
barhanc Oct 9, 2026
005a09d
update
barhanc Oct 9, 2026
24ce001
test(llm): drive the chat session through the TypeScript LLM runner
barhanc Oct 9, 2026
65a31a4
feat(nlp-app): add a backend picker to the LLM screen
barhanc Oct 9, 2026
58e92e0
chore(core): attribute the input shape-check TODO in model.cpp
barhanc Oct 9, 2026
a41def8
update
barhanc Oct 9, 2026
73bb49d
test(cpp): cover the native LLM sampler
barhanc Oct 9, 2026
f99ffe5
style
barhanc Oct 9, 2026
e4669fe
docs(llm): document the TypeScript LLM runner and the native sampler
barhanc Oct 9, 2026
8d7feb6
feat(models): restore the LLM variants pending a re-export
barhanc Oct 9, 2026
f3c231b
fix(models): point LLM tokenizers back at the shared repo-root files
barhanc Oct 9, 2026
a34884f
style
barhanc Oct 9, 2026
d6f3119
feat(models): point the LLMs at the re-exported scratch builds
barhanc Oct 9, 2026
7b2c7c8
feat(nlp-app): list every LLM variant on the LLM screen
barhanc Oct 9, 2026
bfc91c1
fix
barhanc Oct 9, 2026
3f5196c
fix
barhanc Oct 9, 2026
ea5b543
update
barhanc Oct 10, 2026
45c9c0f
update
barhanc Oct 12, 2026
2b35466
refactor(llm): use tensor array destructuring pattern in multimodal r…
barhanc Oct 12, 2026
e74754b
feat(nlp-app): set tailored generation configs for each LLM model family
barhanc Oct 12, 2026
253391a
refactor(llm): align runner reset bounds and clean up gemma runner TODOs
barhanc Oct 12, 2026
6f03765
fix: remove llmBrand
barhanc Oct 13, 2026
7ad4a1d
test(llm): add tests for multimodal runner, variant dispatch, sampler…
barhanc Oct 13, 2026
4462dac
refactor(llm): initialize prefill totStats to zero stats and omit non…
barhanc Oct 13, 2026
1d4b3b4
fix(core): reject non-finite numbers in double and float conversions
barhanc Oct 13, 2026
893c08f
feat(llm): add penaltyLastN window and seed to sampler with full-chai…
barhanc Oct 13, 2026
6d84a23
fix(build): run prepare before workspace typecheck
barhanc Oct 13, 2026
5312189
fix(llm): prevent EOS tokens reaching onToken and fix Hammer/Llama to…
barhanc Oct 13, 2026
5b55ea8
feat(llm): support Gemma Omni model and audio prefill chunking
barhanc Oct 13, 2026
c75e696
test(llm): add test suite for createLLMGemmaRunner and remove TODO in…
barhanc Oct 13, 2026
4376330
test(llm): allow arbitrary shapes in sample and test flattening
barhanc Oct 13, 2026
87fc502
fix
barhanc Oct 13, 2026
a32fce1
fix(llm): mark reset and getKVCacheState as worklet in runners
barhanc Oct 14, 2026
1dbf87f
feat(llm): add Gemma Omni Vulkan model variant
barhanc Oct 14, 2026
37d28ac
feat(models): wire Gemma 4 Omni E2B MLX variant in models registry an…
barhanc Oct 14, 2026
a9eddf1
feat(models): update Qwen 3 0.6B MLX variant to int8 in models regist…
barhanc Oct 14, 2026
f20cb48
feat(models): rename Gemma MLX variant to gemma_4 in models registry
barhanc Oct 14, 2026
d54f17f
fix(llm): match HF Gemma PLE for media prefill and drop padded audio …
barhanc Oct 15, 2026
4b9a932
chore(nlp): update react-native-audio-api to 0.13.6
barhanc Oct 15, 2026
bce7bb9
fix(llm): initialize top-k heap for constrained sampling
barhanc Oct 18, 2026
aab6fd3
feat(nlp): add constrained decoding Potion Lab demo
barhanc Oct 18, 2026
3cd3bc2
fix(nlp): explain constrained decoding validation results
barhanc Oct 18, 2026
966fb0e
docs: await createLLMRunner directly in generation example
barhanc Oct 20, 2026
0b07db7
fix: reject overlapping LLM chat session turns
barhanc Oct 20, 2026
2534cd2
fix: preserve split UTF-8 characters in LLM streaming
barhanc Oct 20, 2026
a715834
docs: clarify local model filesystem path requirements
barhanc Oct 21, 2026
2711a7e
fix: roll back cancelled LLM chat turns
barhanc Oct 21, 2026
bc79589
docs: clarify LLM chat rollback behavior
barhanc Oct 21, 2026
de5d4d1
feat: add LFM2.5 2.6B models and example tool calling
barhanc Oct 21, 2026
f88bf5a
docs: simplify LFM2.5 2.6B model description
barhanc Oct 21, 2026
4fdaf13
fix: pass chat special tokens and advance generation seeds
barhanc Oct 21, 2026
e7cebc1
feat: support configurable LLM prefill length
barhanc Oct 22, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions .cspell-wordlist.txt
Original file line number Diff line number Diff line change
Expand Up @@ -400,6 +400,7 @@ skiplist
skiplisted
startoftext
maxsim
multinomial
adb
benchmarkable
benchprobe
Expand Down
7 changes: 7 additions & 0 deletions apps/nlp/app/_layout.tsx
Original file line number Diff line number Diff line change
Expand Up @@ -55,6 +55,13 @@ export default function Layout() {
title: 'LLM',
}}
/>
<Drawer.Screen
name="constrained-decoding/index"
options={{
drawerLabel: 'Potion Lab',
title: 'Constrained Decoding',
}}
/>
</Drawer>
);
}
Loading
Loading