Spaces:

WaveCut
/

Bonsai-Chat-WebGPU

Running

WaveCut commited on 1 day ago

Commit

20134b6

verified ·

1 Parent(s): 3fd9ca4

Release v1.1.4 manifest progress fix

Correct manifest progress accounting and document verified Safari and Firefox gates; Windows hardware gates remain open.

Files changed (5) hide show

README.md CHANGED Viewed

@@ -58,9 +58,13 @@ shard from byte zero; the UI does not claim resumable Range downloads.
 27B is intentionally gated. It is unavailable when the adapter's
 `maxStorageBufferBindingSize`, `shader-f16`, browser quota, or all-WebGPU graph
 requirement does not pass. Chrome on Apple Silicon is the validated release
-path. Edge/Windows, Safari, and Firefox remain explicit release-matrix gaps;
-dense CPU fallback is offered only when that browser exposes the required
-cross-origin-isolated WASM features.
 ## Growth benchmark

 27B is intentionally gated. It is unavailable when the adapter's
 `maxStorageBufferBindingSize`, `shader-f16`, browser quota, or all-WebGPU graph
 requirement does not pass. Chrome on Apple Silicon is the validated release
+path. Safari 26.5 on Apple Silicon has also passed a real foreground cold 27B
+load through compat/Asyncify plus explicit OPFS eviction. Firefox gating was
+validated on macOS: it selected compat, offered CPU-WASM for dense tiers, and
+refused 27B without WebGPU; Firefox model generation was not exercised.
+Windows Chrome/Edge on a discrete GPU and a physical low-limit iGPU laptop
+remain explicit release-matrix gaps. Dense CPU fallback is offered only when
+the browser exposes the required cross-origin-isolated WASM features.
 ## Growth benchmark

dist/assets/{index-DIzN0g8r.js → index-Dn6W9Rfz.js} RENAMED Viewed

The diff for this file is too large to render. See raw diff

dist/index.html CHANGED Viewed

@@ -10,7 +10,7 @@
       content="Bonsai local inference field station — private WebGPU language models in your browser."
     />
     <title>Bonsai Field Station</title>
-    <script type="module" crossorigin src="/assets/index-DIzN0g8r.js"></script>
     <link rel="stylesheet" crossorigin href="/assets/index-C-2tm91j.css">
   </head>
   <body>

       content="Bonsai local inference field station — private WebGPU language models in your browser."
     />
     <title>Bonsai Field Station</title>
+    <script type="module" crossorigin src="/assets/index-Dn6W9Rfz.js"></script>
     <link rel="stylesheet" crossorigin href="/assets/index-C-2tm91j.css">
   </head>
   <body>

src/app/App.test.ts CHANGED Viewed

@@ -1,6 +1,12 @@
 import { describe, expect, it } from 'vitest';
 import { EngineClientError } from '../engine';
-import { DEFAULT_MODEL_ID, reconcileCompletionTokens, reconcileTotalTokens, requiresModelReload } from './App';
 function engineError(code: string): EngineClientError {
   return new EngineClientError({ code, message: code });
@@ -26,6 +32,14 @@ describe('App release defaults', () => {
   });
 });
 describe('App streamed completion telemetry', () => {
   it('uses observed stream events when final engine usage is incomplete', () => {
     expect(reconcileCompletionTokens(5, 64)).toBe(64);

 import { describe, expect, it } from 'vitest';
 import { EngineClientError } from '../engine';
+import {
+  DEFAULT_MODEL_ID,
+  reconcileCompletionTokens,
+  reconcileTotalTokens,
+  reportsModelWeightProgress,
+  requiresModelReload,
+} from './App';
 function engineError(code: string): EngineClientError {
   return new EngineClientError({ code, message: code });
   });
 });
+describe('App model-load progress', () => {
+  it('does not turn pinned manifest metadata reads into model-weight download state', () => {
+    expect(reportsModelWeightProgress('manifest')).toBe(false);
+    expect(reportsModelWeightProgress('download')).toBe(true);
+    expect(reportsModelWeightProgress('load')).toBe(true);
+  });
+});
 describe('App streamed completion telemetry', () => {
   it('uses observed stream events when final engine usage is incomplete', () => {
     expect(reconcileCompletionTokens(5, 64)).toBe(64);

src/app/App.tsx CHANGED Viewed

@@ -8,6 +8,7 @@ import {
   type EngineCapabilities,
   type EngineChatMessage,
   type EngineChatToolCall,
   type ManifestModelV2,
   type ModelManifestV2,
   type ModelTierId,
@@ -149,6 +150,12 @@ export function requiresModelReload(error: unknown): error is EngineClientError
   ].includes(error.code);
 }
 export function reconcileCompletionTokens(
   engineCompletionTokens: number | undefined,
   streamedTokenEvents: number,
@@ -475,6 +482,10 @@ export function App() {
       }, {
         signal: controller.signal,
         onProgress: (progress) => {
           const received = Math.min(progress.loadedBytes, model.downloadBytes);
           const shard = progress.shardIndex === null ? '' : ` · shard ${progress.shardIndex + 1}/${progress.shardCount}`;
           setBootStatus(`${progress.phase}${shard} · ${formatBytes(received)} / ${formatBytes(model.downloadBytes)}`);

   type EngineCapabilities,
   type EngineChatMessage,
   type EngineChatToolCall,
+  type EngineEvent,
   type ManifestModelV2,
   type ModelManifestV2,
   type ModelTierId,
   ].includes(error.code);
 }
+export function reportsModelWeightProgress(
+  phase: Extract<EngineEvent, { event: 'progress' }>['phase'],
+): boolean {
+  return phase !== 'manifest';
+}
 export function reconcileCompletionTokens(
   engineCompletionTokens: number | undefined,
   streamedTokenEvents: number,
       }, {
         signal: controller.signal,
         onProgress: (progress) => {
+          if (!reportsModelWeightProgress(progress.phase)) {
+            setBootStatus('Reading pinned model manifest…');
+            return;
+          }
           const received = Math.min(progress.loadedBytes, model.downloadBytes);
           const shard = progress.shardIndex === null ? '' : ` · shard ${progress.shardIndex + 1}/${progress.shardCount}`;
           setBootStatus(`${progress.phase}${shard} · ${formatBytes(received)} / ${formatBytes(model.downloadBytes)}`);