BarunAction-35M

A 35-million-parameter local compiler for typed personal actions.

BarunAction-35M turns a natural-language request, explicit tool schemas, context, and a reference time into strict Action IR JSON. It is small enough for local research, hash-verifies its own checkpoint, validates every proposal deterministically, and never invokes a real tool.

Try the live interactive demo · Download the model · Read the model card · Reproduce the benchmark

Release highlights

	BarunAction-35M candidate-v2
Parameters	35,072,768
Model size	140.3 MB float; 57.5 MB Darwin ARM64 int8 payload
Output	Strict, typed Action IR JSON
Mobile Actions development exact match	602/756 (79.63%)
Matched Qwen2.5-0.5B-Instruct comparison	Retains 90.80% of exact-match rate at 14.09× fewer parameters
Runtime boundary	Proposals only; external execution is always disabled
License	Apache-2.0 code and released weights

The comparison model is exactly 494,032,768 parameters and scored 663/756 (87.70%). BarunAction trails it by 61 rows, or 8.07 percentage points. These are strict results on a grouped 756-row Mobile Actions development split derived from public training rows—not the official evaluation split or a broad safety benchmark.

See it work

The Hugging Face Space offers curated calendar, contacts, settings, email, maps, and flashlight examples. It shows:

the exact prompt contract supplied to the compact model;
the generated Action IR;
deterministic parse and schema validation;
a readable explanation of the proposed action; and
an explicit safety summary showing that no real side effect was executed.

The Space and local demo never connect to a calendar, contact book, mail service, map provider, or device API.

Quick start

git clone https://github.com/harrrshall/barunaction-35m.git
cd barunaction-35m
python -m venv .venv
source .venv/bin/activate
pip install -e .

# Weight-free contract and sandbox demonstration
barunaction demo

# Download candidate-v2 from Hugging Face and verify every required SHA-256
barunaction download --output ./models/BarunAction-35M

# Verify again at any time
barunaction verify --checkpoint ./models/BarunAction-35M

# Run deterministic CPU inference
barunaction infer \
  --checkpoint ./models/BarunAction-35M \
  --tools examples/barunaction_tools.example.json \
  --context examples/barunaction_empty_context.example.json \
  --now 2026-08-03T20:00:00+05:30 \
  --request "Turn on the flashlight" \
  --device cpu

Every successful result still carries policy.execution_permitted: false. A proposal becomes a real action only if an application supplies its own authentication, authorization, confirmation, rate limiting, and tool execution layer outside this package.

Python API

import json

from barunaction import BarunActionCompiler, download_candidate_checkpoint

downloaded = download_candidate_checkpoint("./models/BarunAction-35M")
compiler = BarunActionCompiler(downloaded.path, device="cpu")

with open("examples/barunaction_tools.example.json", encoding="utf-8") as source:
    tools = json.load(source)

result = compiler.infer(
    request="Turn on the flashlight",
    tool_schemas=tools,
    context={},
    now="2026-08-03T20:00:00+05:30",
)
print(json.dumps(result.to_dict(), indent=2))

Inputs are rejected unless the request is non-empty, tool declarations are strict and non-duplicated, context is a JSON object, and now is timezone-aware. Outputs are parsed as one complete JSON object without extraction, repair, coercion, or implicit defaults.

What the released model learned

Candidate-v2 was post-trained for one epoch with full-parameter, response-only supervised fine-tuning from the exact harrrshall/BarunLM-35M base checkpoint. It saw 7,937 derived examples once and completed 126 optimizer steps. It used no synthetic teacher data, distillation, preference optimization, DPO, or reinforcement learning.

The released training population covers seven Mobile Actions tools:

create_calendar_event
create_contact
open_wifi_settings
send_email
show_map
turn_off_flashlight
turn_on_flashlight

See Training for the exact recipe and Data card for revisions, derivation, licenses, hashes, and redistribution boundaries.

Benchmark result and terminology

The primary metric is strict Action IR AST exact match. A prediction must be one valid JSON object, satisfy the supplied tool schema, match the decision and call mode, preserve ordered calls for SINGLE/SERIAL, and match typed arguments exactly. PARALLEL calls are compared as an order-independent multiset. Missing outputs, parse failures, schema failures, truncations, and generation failures remain incorrect.

Model	Parameters	Parse valid	Schema valid	Strict AST exact
BarunAction-35M candidate-v2	35,072,768	756/756	755/756	602/756 (79.63%)
Qwen2.5-0.5B-Instruct	494,032,768	755/756	754/756	663/756 (87.70%)

The comparison used identical 7,937 training IDs, 756 development IDs, semantic labels, one pass, effective batch 63, 126 optimizer steps, seed 17, final-checkpoint selection, and unconstrained deterministic greedy decoding. Model-native tokenizers/templates and learning rates differed, and BarunAction had an earlier development sweep. This is therefore a one-seed matched semantic-adaptation comparison, not a universal model ordering.

The source is google/mobile-actions revision e920309bc2acbc2e99a5e3201cf37df2b9fd9151 (CC BY 4.0). The reported 756 rows are grouped public development data and contain only CALL cases. The separate 961-row official evaluation split was never read or materialized.

Full definitions, evaluator logic, exact model revisions, data hashes, decoding settings, and curated sample-level outcome evidence are in Benchmark and Evaluation. The executable scorer is src/barunlm/evaluation/mobile_actions.py.

Reproduce

The public repository separates four layers so results are inspectable without internal lab state:

src/                 model, training, inference, schema, scorer, quantization
configs/             released candidate and matched-baseline contracts
benchmarks/          curated manifests, aggregates, paired outcomes, verification
docs/                model, data, training, evaluation, retrieval, deployment cards
space/               public Gradio demonstration

Common verification commands:

# Product runtime and scorer tests
pytest -q

# Style and formatting
ruff check .
ruff format --check .

# Rebuild the grouped public train/dev manifests from the pinned source.
# Official evaluation rows remain opaque and are never parsed or materialized.
python -m barunlm.datasets.mobile_actions \
  --download-pinned \
  --tokenizer-json ./models/BarunAction-35M/tokenizer.json \
  --tokenizer-revision candidate-v2 \
  --output-dir ./build/mobile-actions

# Re-score the published BarunAction predictions without generating new outputs
barunaction score-mobile \
  --manifest ./build/mobile-actions/dev.jsonl \
  --predictions benchmarks/evidence/barunaction-predictions.jsonl \
  --output ./reproduced-score

No raw training dataset or sealed evaluation label is redistributed. The repository supplies pinned source revisions, derivation code, membership and content hashes, licenses, evaluator code, aggregate results, and curated sample-level comparison outcomes. See Release evidence.

Model files

The canonical Hugging Face mirror contains:

File	SHA-256
`model.safetensors`	`fdb95ccf58a095e0d321be998924318b35ee59a334f6dd97d8726d2cf80021d3`
`tokenizer.json`	`70ded9605fccd09c2340ca7e225361eab0ae8b4dbbb0d6e26343ab5183979db6`
`barun_config.json`	`9b3a1d71baa95a198744d250f9629231738d942570b8685c44307fd83dd33565`
`checkpoint_manifest.json`	`c743ab7c4d33ae75c6b0aa4547458a961b92766da8fcf85fd148fda2ebb5530a`

The source artifacts remain immutable public W&B v0 releases. Hugging Face is a byte-identical, freshly verified mirror. See Retrieval for both locations.

Scope and safety

BarunAction-35M is a compact proposal compiler, not an autonomous assistant. Its public development set contains no safety denominator and does not establish performance on ambiguity, unsafe requests, out-of-scope requests, unseen schemas, or broad function calling. Product integrations must validate identity and permissions outside the model and must never treat valid JSON as authorization.

The included simulator is intentionally in-memory only. It has no network client and no real tool handlers. This makes the demo useful for understanding the contract without risking a message, calendar edit, or device change.

Citation

@software{harrrshall2026barunaction35m,
  author  = {Harrrshall},
  title   = {BarunAction-35M: A Compact Typed Personal-Action Compiler},
  year    = {2026},
  version = {candidate-v2},
  url     = {https://github.com/harrrshall/barunaction-35m}
}

License

Source code and released model weights are licensed under Apache License 2.0. Mobile Actions is licensed CC BY 4.0 and retains its upstream attribution and terms. See NOTICE and the data card.