doppler-gpu 0.4.4 → 0.4.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -1,8 +1,11 @@
1
1
  # doppler-gpu
2
2
 
3
- Browser-native inference on raw WebGPU. Pure JS + WGSL.
3
+ WebGPU inference runtime for browser, Node, Bun, CLI, and local server use.
4
+ Doppler runs JavaScript orchestration over WGSL kernels and loads models from
5
+ RDRR manifests.
4
6
 
5
7
  **[Try the live demo](https://d4da.com/doppler)** | **[npm](https://www.npmjs.com/package/doppler-gpu)** | **[docs](https://github.com/clocksmith/doppler/blob/main/docs/INDEX.md)**
8
+
6
9
  Broader model status and compare evidence live in the support and release
7
10
  matrices. See the
8
11
  [benchmark methodology](https://github.com/clocksmith/doppler/blob/main/docs/benchmark-methodology.md)
@@ -17,9 +20,9 @@ for the receipt contract and disclosure rules.
17
20
  - A runtime that keeps kernel paths, dtype policy, and benchmark contracts
18
21
  visible in JSON, JavaScript, and WGSL.
19
22
 
20
- ## Main goals
23
+ ## Goals
21
24
 
22
- Doppler's mainline work is organized around three goals:
25
+ Doppler's mainline work is organized around three concrete surfaces:
23
26
 
24
27
  1. Make local WebGPU inference a real product surface across the hosted browser
25
28
  demo, `npx doppler-gpu`, root API, CLI, Node/Bun, and OpenAI-compatible
@@ -31,9 +34,11 @@ Doppler's mainline work is organized around three goals:
31
34
  benchmark artifacts, command parity, explicit kernel paths, and fail-closed
32
35
  unsupported paths.
33
36
 
34
- See [docs/goals.md](docs/goals.md) for the compact product and technical
35
- contract behind those goals. The current completion gate is
36
- `src/config/goal-completion-matrix.json`, checked by `npm run goals:check`.
37
+ See
38
+ [docs/goals.md](https://github.com/clocksmith/doppler/blob/main/docs/goals.md)
39
+ for the product and technical contract behind those goals. The current
40
+ completion gate is `src/config/goal-completion-matrix.json`, checked by
41
+ `npm run goals:check`.
37
42
 
38
43
  ## How it works
39
44
 
@@ -128,13 +133,20 @@ surface-specific behavior.
128
133
 
129
134
  ## Benchmark evidence
130
135
 
131
- The release matrix lists checked benchmark fixtures with hardware and backend,
132
- including Apple Metal and AMD Vulkan rows. The README chart uses one reporting
133
- contract across backends: each row states backend, surface, comparator, metric
134
- direction, result, claim state, and evidence path. Metal and Vulkan rows are
135
- separate evidence lanes and should only be compared within a row.
136
+ The npm package includes the runtime, config contracts, schemas, and quickstart
137
+ surface. It does not bundle the full benchmark result tree, local hardware
138
+ artifacts, browser run outputs, or release evidence reports.
139
+
140
+ The GitHub release matrix lists checked benchmark fixtures with hardware and
141
+ backend, including Apple Metal and AMD Vulkan rows. Each row states backend,
142
+ surface, comparator, metric direction, result, claim state, and evidence path.
143
+ Metal and Vulkan rows are separate evidence lanes and should only be compared
144
+ within a row.
136
145
 
137
- ![Doppler backend evidence summary](./benchmarks/vendors/results/doppler-backend-evidence-summary.svg)
146
+ - Benchmark methodology:
147
+ [docs/benchmark-methodology.md](https://github.com/clocksmith/doppler/blob/main/docs/benchmark-methodology.md)
148
+ - Backend evidence summary:
149
+ [benchmarks/vendors/results/doppler-backend-evidence-summary.svg](https://github.com/clocksmith/doppler/blob/main/benchmarks/vendors/results/doppler-backend-evidence-summary.svg)
138
150
 
139
151
  ## Quickstart-supported models
140
152
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "doppler-gpu",
3
- "version": "0.4.4",
3
+ "version": "0.4.5",
4
4
  "description": "Browser-native WebGPU inference engine for local intent and inference loops",
5
5
  "main": "src/index.js",
6
6
  "types": "src/index.d.ts",
@@ -91,8 +91,12 @@ function requireSessionDecodeLoopStopCheckMode(value, modelId) {
91
91
  }
92
92
 
93
93
  function resolveDecodeLoopRuntimeSession(runtimeConfig, runtimeOverrides, useExplicitRuntimeOverrides) {
94
+ const runtimeSession = runtimeConfig?.inference?.session ?? null;
95
+ if (runtimeConfig?.inference?.kernelPathSource === 'execution-v1') {
96
+ return runtimeSession;
97
+ }
94
98
  if (!useExplicitRuntimeOverrides) {
95
- return runtimeConfig?.inference?.session ?? null;
99
+ return runtimeSession;
96
100
  }
97
101
  const inferenceOverrides = runtimeOverrides?.inference;
98
102
  if (!inferenceOverrides || typeof inferenceOverrides !== 'object' || Array.isArray(inferenceOverrides)) {
package/src/version.js CHANGED
@@ -1,2 +1,2 @@
1
- export const DOPPLER_VERSION = '0.4.3';
1
+ export const DOPPLER_VERSION = '0.4.5';
2
2
  export const DOPPLER_PROVIDER_VERSION = DOPPLER_VERSION;