GGUF Q4_K_M -tuki 3B-mallille: kvantisoidtu versio (~1.9 GB) mahtuu selaimeen

Safetensors-muotoinen 3B (~6.2 GB) aiheutti WASM capacity overflow.
Nyt käytetään candle quantized_qwen2 -moduulia GGUF-tiedoston lataamiseen.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-04-06 13:54:23 +03:00
parent 99b011e399
commit d18314bfc8
2 changed files with 48 additions and 34 deletions

View File

@@ -2229,7 +2229,7 @@ Write the corrected code.`;
// Mallikatalogista valinta numerolla tai nimellä
const loadModels = [
{ id: '1', key: '05b', name: 'Qwen2.5-Coder:0.5B', size: '~990 MB', coderSize: '05b' },
{ id: '2', key: '3b', name: 'Qwen2.5-Coder:3B', size: '~6.2 GB', coderSize: '3b' },
{ id: '2', key: '3b', name: 'Qwen2.5-Coder:3B Q4', size: '~1.9 GB', coderSize: '3b' },
];
if (!arg) {
// Näytetään lista
@@ -2268,7 +2268,7 @@ Write the corrected code.`;
if (sub === 'models') {
termLog(' Käytettävissä olevat mallit:', '#c9d1d9');
termLog(' <span style="color:#58a6ff">1</span> qwen-coder Qwen2.5-Coder:0.5B <span style="color:#8b949e">~990 MB | koodin generointi</span>');
termLog(' <span style="color:#58a6ff">2</span> qwen-coder-3b Qwen2.5-Coder:3B <span style="color:#8b949e">~6.2 GB | parempi koodinlaatu</span>');
termLog(' <span style="color:#58a6ff">2</span> qwen-coder-3b Qwen2.5-Coder:3B Q4 <span style="color:#8b949e">~1.9 GB | kvantisoidtu, parempi laatu</span>');
termLog(' <span style="color:#58a6ff">3</span> smollm-135m SmolLM 135M <span style="color:#8b949e">~270 MB | kevyt, nopea</span>');
termLog(' <span style="color:#58a6ff">4</span> qwen-05b Qwen2.5:0.5B <span style="color:#8b949e">~990 MB | yleismalli</span>');
termLog(' <span style="color:#58a6ff">5</span> phi3-mini Phi-3 Mini <span style="color:#8b949e">~2.2 GB | Microsoftin malli</span>');