Winnow E4B

• License: Apache-2.0
• Original model: EldanRing/Winnow-E4B
• Base model: google/gemma-4-E4B-it
• Served modalities: Text and image input; choice, score, and noul output.

Description

Winnow E4B is EldanRing's merged Gemma 4 E4B fine-tune for typed decisions, chat, and image input. This registry entry serves typed decisions through ggmlc-custom-decider with protocol=winnow. The engine uses Winnow's Gemma prompt and next-token option-label probabilities, not the Decider prompt. Each question is evaluated independently; shared-prefix caching is not implemented.

Run

self serve winnow:4b

Use the Decision API at /v1/systemone. Image input uses the request's images array; the current adapter accepts one image per request. Chat, audio, and video are not exposed by this entry.

Quantizations and Files

The default and only registered variant is q8, using the author's merged gguf/Winnow-E4B-Q8_0.gguf (8,005,437,472 bytes) and matching F16 vision projector gguf/mmproj-Winnow-E4B.gguf (990,372,672 bytes). Both files are GGUF, pinned to revision ed891438b340bb60e2962e335fe100b2796c2124 with SHA-256 checks in the registry. No separate base model, LoRA adapter, or speculative assistant is needed.

Settings and Limits

Startup context is 8,192 tokens; the upstream Q8 runtime profile has tested 65,536 positions, including image tokens. Override with --set context_size=65536 when memory permits. Decisions do not generate text. Model download size excludes KV cache and runtime overhead.

The upstream fitted Q8 decision temperature of 1.2574172017327816 applies to option-logit scoring. Up to 64 options are supported. On 2026-10-13, self check passed all CPU choice-routing, option-order, true/false, and score-ordering probes. Image-decision tests also passed red/green/blue classification, malformed-image rejection, and rejection of requests with more than one image. These smoke tests are not a benchmark or an upstream numerical-parity check.

Quality benchmarks

Decision Index: #87 (38.73)

EldanRing/Winnow-E4B

Published result for the named upstream model and evaluated configuration; checkpoint revision and local quantization equivalence are not guaranteed.

Evaluated: winnow-e4b-q8 (Q8_0)

Edition: Decision Index 0.3.2

Checked 2026-10-13

Benchmarks

No submitted benchmark reports yet.