Downloads · 30 days
16
36% of all-time downloads
momo-watch
MadhavOberoi1121/momo-watch
momo-watch is a machine learning model from MadhavOberoi1121. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for gguf. The card lists the license as apache-2.0.
The model behind Momo, a voice assistant that runs entirely on Apple Watch with llama.cpp. One model does both jobs: it turns a spoken request into a tool call (timers, reminders, calendar, weather, calls, messages, n…
All-time downloads
44
Public
Repo size
271 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.gguf271 MB · 100%
At a glance
- Library
- gguf
- License
- apache-2.0
- Access
- Public
- Created
- Sep 24, 2026
- Updated
- Sep 24, 2026
- SHA
- ba43b6bb
Base models
Model card
From the Hugging Face model README
Momo 360M
The model behind Momo, a voice assistant that runs entirely on Apple Watch with llama.cpp. One model does both jobs: it turns a spoken request into a tool call (timers, reminders, calendar, weather, calls, messages, notes, workouts, health, directions, web search) and it writes short answers, code and replies.
- Base: HuggingFaceTB/SmolLM2-360M-Instruct (Apache-2.0)
- Changes: full fine-tune (MLX) on synthetic Momo tool-call data plus self-distilled replies from the base model under Momo's concise prompt, so it keeps its writing ability. Quantized to Q4_K_M GGUF.
- Runs on: Apple Watch SE (2nd gen) and later, watchOS 11+, via llama.cpp (arm64_32).
Prompt formats (ChatML)
Tool calls: a system turn listing the tools, then the user's words. The model replies with
<tool_call>{"name": ..., "arguments": {...}}</tool_call>.
Answers: a system turn asking for concise replies, then the question (optionally with reference text).
License
Apache-2.0, as the base model. This is a modified version of SmolLM2-360M-Instruct (fine-tuned and quantized).
- Library
- gguf
- License
- apache-2.0
- Languages
- en
- Created
- Sep 24, 2026
- Updated
- Sep 24, 2026