volvoxai 0.6.0 → 0.7.0
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 +3 -3
- package/dist/{0.6.0 → 0.7.0}/volvoxai.js +2385 -601
- package/dist/{0.6.0 → 0.7.0}/volvoxai.lite.js +1285 -573
- package/dist/0.7.0/volvoxai.lite.min.js +2 -0
- package/dist/0.7.0/volvoxai.lite.wasm +0 -0
- package/dist/0.7.0/volvoxai.min.js +2 -0
- package/dist/0.7.0/volvoxai.wasm +0 -0
- package/package.json +15 -12
- package/dist/0.6.0/volvoxai.lite.min.js +0 -2
- package/dist/0.6.0/volvoxai.lite.wasm +0 -0
- package/dist/0.6.0/volvoxai.min.js +0 -2
- package/dist/0.6.0/volvoxai.wasm +0 -0
package/README.md
CHANGED
|
@@ -165,7 +165,7 @@ batching is useful when the graph preserves each request's independence.
|
|
|
165
165
|
|
|
166
166
|
For text generation, prefill a context with a prompt and then advance it one
|
|
167
167
|
step at a time. Its KV cache retains previous attention state. Paged caches can
|
|
168
|
-
share prefixes and retire individual
|
|
168
|
+
share prefixes and retire individual slots. `DecodeGenerate` handles a supported
|
|
169
169
|
fixed-count greedy feedback loop; sampling, stop conditions, and task policy
|
|
170
170
|
remain application decisions. See [scheduling and dynamic batching](docs/scheduling-and-dynamic-batching-design.md).
|
|
171
171
|
|
|
@@ -220,7 +220,7 @@ and a PTQ command-line workflow.
|
|
|
220
220
|
|
|
221
221
|
```sh
|
|
222
222
|
make build_wheel
|
|
223
|
-
python3 -m pip install dist/python/0.
|
|
223
|
+
python3 -m pip install dist/python/0.7.0/*.whl
|
|
224
224
|
```
|
|
225
225
|
|
|
226
226
|
For an exported model with one input, `InferenceSession` handles loading,
|
|
@@ -252,7 +252,7 @@ integration and engine-development guides. Useful starting points:
|
|
|
252
252
|
- [Quickstart](docs/quickstart.md) and [textbook](docs/textbook/README.md)
|
|
253
253
|
- [Model format](docs/model-format.md) and [operator support](docs/operation_list.md)
|
|
254
254
|
- [Architecture](ARCHITECTURE.md) and [backend development](docs/backend-sdk.md)
|
|
255
|
-
- [Testing](docs/testing.md)
|
|
255
|
+
- [Testing](docs/testing.md), [profiling](docs/profiling.md) and [node debugging](docs/debugging.md)
|
|
256
256
|
|
|
257
257
|
The schema in [volvoxai.proto](proto/volvoxai.proto) defines the shared API for
|
|
258
258
|
applications and AI agents. For field-level lookup and programmatic discovery,
|