doppler-gpu 0.2.1 → 0.4.2

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.
Files changed (65) hide show
  1. package/CHANGELOG.md +8 -0
  2. package/README.md +9 -1
  3. package/assets/doppler.svg +1601 -0
  4. package/package.json +6 -5
  5. package/src/cli/doppler-quickstart.js +16 -2
  6. package/src/cli/doppler-serve.js +0 -0
  7. package/src/client/doppler-registry.json +17 -1
  8. package/src/config/conversion/gemma4/gemma-4-e2b-it-q4k-ehf16-af32.json +1 -1
  9. package/src/config/kernels/registry.json +73 -0
  10. package/src/config/runtime/experiments/bench/gemma4-bench-q4k.json +32 -0
  11. package/src/config/schema/benchmark.schema.d.ts +2 -0
  12. package/src/config/schema/benchmark.schema.js +1 -0
  13. package/src/config/schema/doppler.schema.d.ts +9 -1
  14. package/src/config/schema/doppler.schema.js +8 -1
  15. package/src/config/schema/index.d.ts +2 -0
  16. package/src/config/schema/index.js +1 -0
  17. package/src/config/schema/manifest.schema.d.ts +5 -0
  18. package/src/converter/quantizer.js +5 -4
  19. package/src/formats/rdrr/source-transform-contract.js +390 -0
  20. package/src/formats/rdrr/types.d.ts +8 -1
  21. package/src/formats/tflite/types.d.ts +8 -0
  22. package/src/formats/tflite/types.js +45 -0
  23. package/src/gpu/command-recorder.d.ts +7 -1
  24. package/src/gpu/command-recorder.js +41 -2
  25. package/src/gpu/kernels/activation-static-qdq.d.ts +22 -0
  26. package/src/gpu/kernels/activation-static-qdq.js +69 -0
  27. package/src/gpu/kernels/activation_static_qdq.wgsl +23 -0
  28. package/src/gpu/kernels/index.d.ts +7 -0
  29. package/src/gpu/kernels/index.js +6 -0
  30. package/src/inference/browser-harness-text-helpers.d.ts +1 -0
  31. package/src/inference/browser-harness-text-helpers.js +44 -4
  32. package/src/inference/browser-harness.js +7 -0
  33. package/src/inference/pipelines/context.js +4 -2
  34. package/src/inference/pipelines/text/attention/types.js +9 -4
  35. package/src/inference/pipelines/text/embed.d.ts +1 -0
  36. package/src/inference/pipelines/text/embed.js +10 -4
  37. package/src/inference/pipelines/text/generator-helpers.d.ts +6 -0
  38. package/src/inference/pipelines/text/generator-helpers.js +20 -1
  39. package/src/inference/pipelines/text/generator-steps.js +17 -8
  40. package/src/inference/pipelines/text/generator.js +51 -3
  41. package/src/inference/pipelines/text/init.d.ts +3 -0
  42. package/src/inference/pipelines/text/init.js +24 -3
  43. package/src/inference/pipelines/text/logits/gpu.d.ts +5 -4
  44. package/src/inference/pipelines/text/logits/gpu.js +146 -11
  45. package/src/inference/pipelines/text/per-layer-inputs.js +320 -57
  46. package/src/inference/pipelines/text/probes.js +79 -1
  47. package/src/inference/pipelines/text/stage-names.js +9 -0
  48. package/src/inference/pipelines/text.js +10 -2
  49. package/src/loader/embedding-loader.js +20 -5
  50. package/src/loader/loader-types.d.ts +2 -0
  51. package/src/loader/manifest-config.d.ts +2 -7
  52. package/src/loader/per-layer-input-loader.js +114 -1
  53. package/src/loader/tensors/source-transform.d.ts +15 -0
  54. package/src/loader/tensors/source-transform.js +312 -34
  55. package/src/loader/tensors/tensor-reader.js +28 -2
  56. package/src/storage/artifact-storage-context.d.ts +1 -0
  57. package/src/storage/artifact-storage-context.js +158 -72
  58. package/src/storage/preflight.js +11 -0
  59. package/src/storage/quickstart-downloader.js +15 -4
  60. package/src/tooling/litert-package-runtime.js +369 -14
  61. package/src/tooling/node-source-runtime.js +256 -140
  62. package/src/tooling/source-runtime-bundle.d.ts +25 -0
  63. package/src/tooling/source-runtime-bundle.js +53 -1
  64. package/src/tooling/source-runtime-materializer.js +3 -1
  65. package/src/version.js +1 -1
package/CHANGELOG.md CHANGED
@@ -6,6 +6,14 @@ This changelog is package-facing and release-oriented. Entries before `0.1.7`
6
6
  were retrofitted from package version history, release commits, and release
7
7
  docs so the `0.1.x` line has one conventional npm-visible history surface.
8
8
 
9
+ ## [0.2.2] - 2026-04-13
10
+
11
+ ### Fixed
12
+
13
+ - Switched the README benchmark image to an absolute GitHub raw URL so the npm
14
+ package page renders the current published chart instead of relying on an
15
+ unpublished local asset path.
16
+
9
17
  ## [0.2.1] - 2026-04-13
10
18
 
11
19
  ### Changed
package/README.md CHANGED
@@ -1,10 +1,12 @@
1
1
  # doppler-gpu
2
2
 
3
+ ![Doppler logo](./assets/doppler.svg)
4
+
3
5
  Browser-native inference on raw WebGPU. Pure JS + WGSL.
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)**
6
8
 
7
- [![Published phase-latency comparison across Gemma 4 E2B and Gemma 3 1B](./benchmarks/vendors/results/compare_1b_multi-workload_favorable_phases.svg)](https://github.com/clocksmith/doppler/blob/main/docs/benchmark-methodology.md)
9
+ [![Published phase-latency comparison across Gemma 4 E2B and Gemma 3 1B](https://raw.githubusercontent.com/clocksmith/doppler/main/benchmarks/vendors/results/compare_1b_multi-workload_favorable_phases.svg)](https://github.com/clocksmith/doppler/blob/main/docs/benchmark-methodology.md)
8
10
 
9
11
  This chart is scoped to two published warm compare lanes: Gemma 4 E2B and
10
12
  Gemma 3 1B. It is no longer the old Gemma 3 + LFM snapshot. Broader model
@@ -12,6 +14,11 @@ status and the surrounding compare evidence live in the support and release
12
14
  matrices.
13
15
  See the [benchmark methodology](https://github.com/clocksmith/doppler/blob/main/docs/benchmark-methodology.md).
14
16
 
17
+ ## New in 0.4.2
18
+
19
+ - `Gemma 4 E2B` support for Doppler-native `RDRR` artifacts and direct-source `TFLite` / LiteRT workflows.
20
+ - Experimental direct-source loading for `TFLite`, `GGUF`, and `SafeTensors`, alongside native `RDRR` artifacts.
21
+
15
22
  ## Quick start
16
23
 
17
24
  ### Browser
@@ -110,6 +117,7 @@ These registry IDs resolve to hosted RDRR artifacts automatically from the brows
110
117
  | --- | --- | --- | --- | --- |
111
118
  | Gemma 3 270M IT | `gemma3-270m` | Q4K | 270M | Gemma |
112
119
  | Gemma 3 1B IT | `gemma3-1b` | Q4K | 1B | Gemma |
120
+ | Gemma 4 E2B IT | `gemma4-e2b` | Q4K | E2B | Gemma |
113
121
  | EmbeddingGemma 300M | `embeddinggemma-300m` | Q4K | 300M | Gemma |
114
122
  | Qwen 3.5 0.8B | `qwen3-0.8b` | Q4K | 0.8B | Qwen |
115
123
  | Qwen 3.5 2B | `qwen3-2b` | Q4K | 2B | Qwen |