Instructions to use nakas/kiln-photo-1.5b-q4f16_1-MLC with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLC-LLM
How to use nakas/kiln-photo-1.5b-q4f16_1-MLC with MLC-LLM:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Kiln Photo 1.5B (q4f16_1, MLC / WebLLM)
A LoRA fine-tune of Qwen2.5-1.5B-Instruct that drives
the AI panel of Kiln Photo, a free browser photo editor, entirely in
the browser with WebLLM. It edits the open document through the
panel's tools (get_document, find_commands, run_commands, undo) using a fenced-JSON text tool
protocol, then replies with one sentence saying what it changed.
- Format: MLC
q4f16_1, same layout asmlc-ai/Qwen2.5-1.5B-Instruct-q4f16_1-MLC, so WebLLM's prebuiltQwen2-1.5B-Instruct-q4f16_1_cs1k-webgpu.wasmmodel library runs it. - Training data: 1,494 editing conversations generated by replaying scripted plans through the real editor and keeping only those whose result passed an automatic check (46 hand-written tasks plus every filter and adjustment in the command registry).
- Held-out eval (84 unseen phrasings, scored on the resulting document): 82% for these weights, versus 4% for the untuned base model with the editor's full prompt.
It is specialised for Kiln Photo's tools and system prompt and is not a general assistant.
- Downloads last month
- 106
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support