# Models

Q8_0 GGUF fill-in-the-middle models for the [Local Autocomplete](../extensions/vscode) extension, fine-tuned from
[Qwen2.5-Coder-0.5B](https://huggingface.co/Qwen/Qwen2.5-Coder-0.5B) for **TypeScript, Vue and Node.js**.

| model | description | prompt order | context modes | size |
|---|---|---|---|---|
| `stage-d` | Suffix-first FIM with symbols (incl. namespace imports) + TS errors; better next-line, block closing and multi-line completions | spm | symbols, simple | 531 MB |
| `stage-e` | Next-edit prediction (rewrites a region around the cursor or answers "no edit"); also does FIM, slightly below stage-d | spm | symbols, simple | 531 MB |
| `stage-c` | Suffix-first FIM with repo/symbol-context headers (default) | spm | symbols, simple | 531 MB |
| `stage-b` | Suffix-first FIM without context headers (previous stage) | spm | raw | 531 MB |

The extension reads [`manifest.json`](manifest.json), downloads the selected model once, verifies its SHA-256 and
runs it locally with llama.cpp. The files are stored with Git LFS; download links use Gitea's `/media/` route:

    https://git.harvmaster.com/Harvmaster/autocomplete-model/media/branch/main/models/<file>

## Evaluation (612 held-out samples from private repos the model never saw, suffix-first prompts)

| | exact match | edit similarity | still parses |
|---|---:|---:|---:|
| Qwen2.5-Coder-0.5B (base) | 0.270 | 0.506 | 0.783 |
| Qwen2.5-Coder-1.5B (base) | 0.366 | 0.623 | 0.882 |
| `stage-b` | 0.391 | 0.689 | 0.952 |
| `stage-c` (with symbol context) | 0.397 | 0.681 | 0.954 |
| `stage-c` + token healing (daemon default) | 0.405 | 0.689 | |
| `stage-d` (symbols + TS errors, healed) | **0.426** | **0.700** | 0.954 |
| `stage-e` (symbols + TS errors, healed) | 0.412 | 0.691 | 0.962 |

New behaviours (886 further held-out samples), exact match, stage-c → stage-d: closing a block 0.27 → **0.79**,
next line 0.01 → **0.18**, rest of block 0.08 → 0.11, member access 0.63 → 0.66.

**Next-edit** (`stage-e`, 445 held-out samples): 80% exact region rewrites (edit similarity 0.99) where an edit
was due, and "no edit" on 100% of samples where none was. `stage-d` uses `stage-e` as its next-edit companion
(`nextEditModel` in the manifest): the extension downloads both and the daemon runs it in a second llama-server.

The base models were evaluated in their native prefix-first order (they can't do suffix-first). Suffix-first prompts
let the editor reuse the KV cache while typing: about 2 tokens re-processed per keystroke, ~90–100 ms end-to-end on a
laptop CPU.

## Using the files directly

Any llama.cpp-based runtime works, but the prompt must match the training layout. The
[daemon](../daemon) and [`shared/prompt.ts`](../shared/prompt.ts) are the reference.
Suffix-first (`spm`) order:

    <|fim_prefix|><|fim_suffix|>{suffix}<|fim_middle|>{prefix}   → model generates the middle, then <|endoftext|>

`stage-c` additionally expects a repo-level header: `<|repo_name|>{repo}\n<|file_sep|>{path}\n`, optionally with a
`__symbols__.d.ts` pseudo-file of available symbols before it.

## Training data

Fine-tuned on permissively licensed open-source repositories (below, pinned commits) and on the publisher's own
code (not published). Secrets were scrubbed from all training data; held-out evaluation repos were never trained on.

| repository | commit | license |
|---|---|---|
| [github.com/WiseLibs/better-sqlite3](https://github.com/WiseLibs/better-sqlite3) | `f8e2d54120` | MIT |
| [github.com/CashScript/cashscript](https://github.com/CashScript/cashscript) | `71a243b656` | MIT |
| [gitlab.com/electrum-cash/network](https://gitlab.com/electrum-cash/network) | `c7511725a9` | MIT |
| [github.com/element-plus/element-plus](https://github.com/element-plus/element-plus) | `ad514f79ba` | MIT |
| [github.com/fastify/fastify](https://github.com/fastify/fastify) | `19d5be0daf` | MIT |
| [github.com/h3js/h3](https://github.com/h3js/h3) | `3afe46b170` | MIT |
| [github.com/kysely-org/kysely](https://github.com/kysely-org/kysely) | `4b7287e71b` | MIT |
| [github.com/bitauth/libauth](https://github.com/bitauth/libauth) | `60aec239cc` | MIT |
| [github.com/nuxt/ui](https://github.com/nuxt/ui) | `e8dd339dfd` | MIT |
| [github.com/unjs/ofetch](https://github.com/unjs/ofetch) | `1dbc37fd1c` | MIT |
| [github.com/vuejs/pinia](https://github.com/vuejs/pinia) | `98587ca465` | MIT |
| [github.com/quasarframework/quasar](https://github.com/quasarframework/quasar) | `0fd7ce30e7` | MIT |
| [github.com/unovue/reka-ui](https://github.com/unovue/reka-ui) | `e9230140ed` | MIT |
| [github.com/unovue/shadcn-vue](https://github.com/unovue/shadcn-vue) | `f12a6a098d` | MIT |
| [github.com/vitejs/vite](https://github.com/vitejs/vite) | `10033218d2` | MIT |
| [github.com/vuejs/vitepress](https://github.com/vuejs/vitepress) | `56fa6c9df1` | MIT |
| [github.com/vitest-dev/vitest](https://github.com/vitest-dev/vitest) | `964e1a4404` | MIT |
| [github.com/vuejs/core](https://github.com/vuejs/core) | `4ab865a848` | MIT |
| [github.com/intlify/vue-i18n](https://github.com/intlify/vue-i18n) | `6ef45366e1` | MIT |
| [github.com/vuejs/router](https://github.com/vuejs/router) | `07663eb6be` | MIT |
| [github.com/vbenjs/vue-vben-admin](https://github.com/vbenjs/vue-vben-admin) | `50f4ede309` | MIT |
| [github.com/vueuse/vueuse](https://github.com/vueuse/vueuse) | `efdd69a148` | MIT |
| [github.com/websockets/ws](https://github.com/websockets/ws) | `297202cdae` | MIT |
| [github.com/colinhacks/zod](https://github.com/colinhacks/zod) | `0b216ef674` | MIT |

## License

Apache-2.0 ([LICENSE](LICENSE), [NOTICE](NOTICE)), as a derivative of Qwen2.5-Coder.