shamrock-0-e4b
Gemma 4 E4B fine-tuned for the Oxalis desktop app's
language-learning surfaces (camera translate, text translate, tutor, talk,
dictionary) across Japanese, Korean, and Simplified Chinese. The E4B sibling
of oxalis-ink/shamrock-0-e2b.
Layout
| Path | Contents |
|---|---|
| repo root | Merged fp16 checkpoint |
gguf/ | shamrock-0-e4b-q4_k_m.gguf + mmproj-shamrock-0-e4b-f16.gguf, what the app's llama.cpp runtime loads |
MTP/ | Multi-token-prediction drafter for speculative decoding: mtp-shamrock-0-e4b-Q8_0.gguf (llama.cpp, serve with --spec-type draft-mtp) and the same head in HF format under MTP/assistant/ |
License
shamrock-0-e4b is licensed under CC BY-NC 4.0 (see LICENSE). You can use, share and adapt it for non-commercial purposes, including personal projects and research, as long as you give credit. Commercial use, such as selling the weights, serving them in a paid product or API, or bundling them into a commercial app, is not permitted. For commercial licensing, reach out via oxalis.ink.
When you share this model or anything built from it, credit it as:
shamrock-0-e4b by Oxalis (https://oxalis.ink), licensed under CC BY-NC 4.0
and say whether you changed it.
shamrock-0-e4b is a modified version of Gemma 4 E4B by Google DeepMind: its weights were fine-tuned and merged, then converted to GGUF with an added MTP drafter. Gemma 4 itself is licensed under the Apache License 2.0 (see LICENSE-GEMMA and the Gemma 4 license); the CC BY-NC 4.0 terms cover Oxalis's modifications. See NOTICE.