goodmai commited on
Commit
4d874db
·
verified ·
1 Parent(s): 22e4cb8

Fit calculator: hide OOM model below 32 GB

Browse files
Files changed (1) hide show
  1. index.html +2 -2
index.html CHANGED
@@ -54,13 +54,13 @@ const MODELS = [
54
  { id: "Youssofal/Ternary-Bonsai-2-27B-MTPLX-Optimized-Speed", gb: 7.4, tps: "21 · 27 with --mtp · 31–38 lookup on code", flags: " --mtp" },
55
  { id: "LiquidAI/LFM2.5-24B-A2B-MLX-4bit", gb: 12.6, tps: "~40 (pre-0.8)" },
56
  { id: "Ex0bit/MYTHOS-26B-A4B-PRISM-PRO-DQ-MLX", gb: 14.5, tps: "~14 (pre-0.8)" },
57
- { id: "mlx-community/gemma-4-31b-it-4bit", gb: 17, tps: "OOM on 24 GB" },
58
  ];
59
  const HEADROOM_GB = 4;
60
 
61
  function render() {
62
  const ram = Number(document.getElementById("ram").value);
63
- const rows = MODELS.filter(m => m.gb + HEADROOM_GB <= ram).map(m => {
64
  const cmd = `basalt serve --model ${m.id}${m.flags || ""} --rest`;
65
  return `<tr><td><a href="https://huggingface.co/${m.id}">${m.id}</a></td><td>${m.gb} GB</td><td>${m.tps}</td>` +
66
  `<td><code>${cmd}</code><br><button data-cmd="${cmd}">Copy</button></td></tr>`;
 
54
  { id: "Youssofal/Ternary-Bonsai-2-27B-MTPLX-Optimized-Speed", gb: 7.4, tps: "21 · 27 with --mtp · 31–38 lookup on code", flags: " --mtp" },
55
  { id: "LiquidAI/LFM2.5-24B-A2B-MLX-4bit", gb: 12.6, tps: "~40 (pre-0.8)" },
56
  { id: "Ex0bit/MYTHOS-26B-A4B-PRISM-PRO-DQ-MLX", gb: 14.5, tps: "~14 (pre-0.8)" },
57
+ { id: "mlx-community/gemma-4-31b-it-4bit", gb: 17, minRam: 32, tps: "OOM on 24 GB, needs 32 GB+" },
58
  ];
59
  const HEADROOM_GB = 4;
60
 
61
  function render() {
62
  const ram = Number(document.getElementById("ram").value);
63
+ const rows = MODELS.filter(m => (m.minRam || m.gb + HEADROOM_GB) <= ram).map(m => {
64
  const cmd = `basalt serve --model ${m.id}${m.flags || ""} --rest`;
65
  return `<tr><td><a href="https://huggingface.co/${m.id}">${m.id}</a></td><td>${m.gb} GB</td><td>${m.tps}</td>` +
66
  `<td><code>${cmd}</code><br><button data-cmd="${cmd}">Copy</button></td></tr>`;