Ollaya package for alibiserikbay/JevK5-GGUF

4b/calibration.json

@@ -0,0 +1,9 @@
+{
+  "temperature": [
+    1.22,
+    1.22,
+    1.22
+  ],
+  "temperature_by_options": {},
+  "source": "JevK5 v0.3 (4B) GGUF temperature 1.22 from the JevK5-GGUF card and jevk5_config.json (alibiserikbay/JevK5@c4f7fdb3); knockout_temperature 0.93 applies above 16 options, which jevk5-v1 does not read"
+}

4b/decision.json

@@ -0,0 +1,64 @@
+{
+ "engine": "llama",
+ "family": "jevk5",
+ "layout": "jevk5-v1",
+ "labels": {
+  "strings": [
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "ids": [
+   32,
+   33,
+   34,
+   35,
+   36,
+   37,
+   38,
+   39,
+   40,
+   41,
+   42,
+   43,
+   44,
+   45,
+   46,
+   47
+  ]
+ },
+ "upstream": {
+  "runtime": "https://github.com/allebee/jevk5",
+  "commit": "f944fe37ff1d5ed3830aa4c8d88b7189c8c1268a",
+  "source": "jevk5/prompt.py"
+ },
+ "gguf": {
+  "repo": "alibiserikbay/JevK5-GGUF",
+  "revision": "ec67b0bfce5119a8b11a2cdb430bb43e3fa3e82a",
+  "path": "jevk5-4b-v0.3-Q8_0.gguf",
+  "sha256": "aea433883bc7ed399f2fbd539e53d2eac7caf71a946fe6650995a413979d4a30",
+  "size": 4482402720,
+  "quantization": "Q8_0",
+  "architecture": "qwen35",
+  "url": "https://huggingface.co/alibiserikbay/JevK5-GGUF/resolve/ec67b0bfce5119a8b11a2cdb430bb43e3fa3e82a/jevk5-4b-v0.3-Q8_0.gguf"
+ },
+ "llama": {
+  "n_ctx": 16384,
+  "swa_full": false,
+  "plan": "cold",
+  "build": "b11146"
+ }
+}

README.md

@@ -0,0 +1,43 @@
+---
+license: apache-2.0
+base_model:
+- alibiserikbay/JevK5-GGUF
+tags:
+- ollaya
+- gguf
+- llama.cpp
+- decision-model
+- system-one
+pipeline_tag: text-classification
+---
+
+# jevk5 for Ollaya
+
+[Ollaya](https://github.com/ollaya-dev/ollaya) package of **[alibiserikbay/JevK5-GGUF](https://huggingface.co/alibiserikbay/JevK5-GGUF)** by alibiserikbay.
+Ollaya runs open decision models locally, the way Ollama runs LLMs: typed questions in,
+calibrated answers out, behind a TypeSafe-compatible API.
+
+```sh
+ollaya run jevk5
+```
+
+## What is in this repository
+
+This repository holds only the files Ollaya derives, with no weights. The model is the authors' own GGUF
+file: `ollaya pull` downloads it from their repository, unmodified and pinned to a commit, verifies its
+sha256, and Ollaya runs it on llama.cpp.
+
+| Tag | The authors' GGUF | Files |
+|---|---|---|
+| `jevk5:4b` | [alibiserikbay/JevK5-GGUF@ec67b0b](https://huggingface.co/alibiserikbay/JevK5-GGUF/tree/ec67b0bfce5119a8b11a2cdb430bb43e3fa3e82a) `jevk5-4b-v0.3-Q8_0.gguf` | `4b/decision.json`, `4b/calibration.json` |
+
+Each tag has `decision.json` (the prompt, the option labels Ollaya reads and llama.cpp's settings) and
+`calibration.json` (temperatures).
+
+## Parity
+
+Ollaya's runner matches stock llama-server of the pinned build (b11146) on the same GGUF, CUDA (RTX 4090): 593 questions, every decision the same, option logits within 7.7e-6 and probabilities within 1.6e-6. The prompts are byte-identical to the author's jevk5.prompt.
+
+## License
+
+Same as the upstream model (Apache-2.0). Ollaya itself is Apache-2.0.