Publish google-embeddinggemma-300m-q4k-ehf16-af32 RDRR artifact
Browse files
models/google-embeddinggemma-300m-q4k-ehf16-af32/manifest.json
CHANGED
|
@@ -138,19 +138,46 @@
|
|
| 138 |
"ringStop": 1,
|
| 139 |
"ringStaging": 1,
|
| 140 |
"disableCommandBatching": false
|
| 141 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 142 |
},
|
| 143 |
"execution": {
|
| 144 |
"kernels": {
|
| 145 |
"embed": {
|
| 146 |
"kernel": "gather.wgsl",
|
| 147 |
"entry": "main",
|
| 148 |
-
"digest": "sha256:
|
| 149 |
},
|
| 150 |
"rmsnorm": {
|
| 151 |
"kernel": "rmsnorm.wgsl",
|
| 152 |
"entry": "main",
|
| 153 |
-
"digest": "sha256:
|
| 154 |
},
|
| 155 |
"tiled": {
|
| 156 |
"kernel": "matmul_f16w_f32a.wgsl",
|
|
@@ -160,7 +187,7 @@
|
|
| 160 |
"rope": {
|
| 161 |
"kernel": "rope.wgsl",
|
| 162 |
"entry": "main",
|
| 163 |
-
"digest": "sha256:
|
| 164 |
},
|
| 165 |
"attn_small": {
|
| 166 |
"kernel": "attention_small.wgsl",
|
|
@@ -337,6 +364,9 @@
|
|
| 337 |
"dtypeTransition": "require_cast_step",
|
| 338 |
"unresolvedKernel": "error"
|
| 339 |
}
|
|
|
|
|
|
|
|
|
|
| 340 |
}
|
| 341 |
},
|
| 342 |
"shards": [
|
|
@@ -4280,7 +4310,7 @@
|
|
| 4280 |
"config": "google-embeddinggemma-300m-q4k-ehf16-af32.json"
|
| 4281 |
},
|
| 4282 |
"manifestRefresh": {
|
| 4283 |
-
"at": "2026-04-
|
| 4284 |
"config": "google-embeddinggemma-300m-q4k-ehf16-af32.json"
|
| 4285 |
}
|
| 4286 |
},
|
|
@@ -4290,5 +4320,11 @@
|
|
| 4290 |
"file": "tokenizer.json",
|
| 4291 |
"addBosToken": true,
|
| 4292 |
"addEosToken": true
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4293 |
}
|
| 4294 |
-
}
|
|
|
|
| 138 |
"ringStop": 1,
|
| 139 |
"ringStaging": 1,
|
| 140 |
"disableCommandBatching": false
|
| 141 |
+
},
|
| 142 |
+
"perLayerInputs": {
|
| 143 |
+
"materialization": "auto",
|
| 144 |
+
"rowCache": {
|
| 145 |
+
"mode": "lru",
|
| 146 |
+
"maxRows": 256,
|
| 147 |
+
"maxBytes": 134217728,
|
| 148 |
+
"decodedDtype": "f32"
|
| 149 |
+
},
|
| 150 |
+
"prefetch": {
|
| 151 |
+
"mode": "next_token",
|
| 152 |
+
"rowsAhead": 1
|
| 153 |
+
},
|
| 154 |
+
"gpuUpload": {
|
| 155 |
+
"mode": "per_step_slices",
|
| 156 |
+
"stagingRows": 2
|
| 157 |
+
},
|
| 158 |
+
"hotCache": {
|
| 159 |
+
"mode": "prepared_tokens",
|
| 160 |
+
"maxTokens": 1024,
|
| 161 |
+
"maxBytes": 134217728,
|
| 162 |
+
"outputDtype": "f32"
|
| 163 |
+
}
|
| 164 |
+
},
|
| 165 |
+
"prefillChunkSubmitMode": "sync",
|
| 166 |
+
"useFlashPrefillAttention": false,
|
| 167 |
+
"useWideTileQ4KPrefill": false,
|
| 168 |
+
"retainQ4KMaterialization": false
|
| 169 |
},
|
| 170 |
"execution": {
|
| 171 |
"kernels": {
|
| 172 |
"embed": {
|
| 173 |
"kernel": "gather.wgsl",
|
| 174 |
"entry": "main",
|
| 175 |
+
"digest": "sha256:4b12653c53247b32ebde7f6cf6a989d6248977e3816c761540b990b5f9818cb6"
|
| 176 |
},
|
| 177 |
"rmsnorm": {
|
| 178 |
"kernel": "rmsnorm.wgsl",
|
| 179 |
"entry": "main",
|
| 180 |
+
"digest": "sha256:c723c0300cc8f93396d79d1a274c0a90ccf1eb8ce82c0d778b8cd4659615fe45"
|
| 181 |
},
|
| 182 |
"tiled": {
|
| 183 |
"kernel": "matmul_f16w_f32a.wgsl",
|
|
|
|
| 187 |
"rope": {
|
| 188 |
"kernel": "rope.wgsl",
|
| 189 |
"entry": "main",
|
| 190 |
+
"digest": "sha256:e13c54f261d7ef00474b71e4fe7b3298fb5cc49161f50c7ea1854aa6672ebe71"
|
| 191 |
},
|
| 192 |
"attn_small": {
|
| 193 |
"kernel": "attention_small.wgsl",
|
|
|
|
| 364 |
"dtypeTransition": "require_cast_step",
|
| 365 |
"unresolvedKernel": "error"
|
| 366 |
}
|
| 367 |
+
},
|
| 368 |
+
"largeWeights": {
|
| 369 |
+
"gpuResidentOverrides": null
|
| 370 |
}
|
| 371 |
},
|
| 372 |
"shards": [
|
|
|
|
| 4310 |
"config": "google-embeddinggemma-300m-q4k-ehf16-af32.json"
|
| 4311 |
},
|
| 4312 |
"manifestRefresh": {
|
| 4313 |
+
"at": "2026-04-10T23:43:07.897Z",
|
| 4314 |
"config": "google-embeddinggemma-300m-q4k-ehf16-af32.json"
|
| 4315 |
}
|
| 4316 |
},
|
|
|
|
| 4320 |
"file": "tokenizer.json",
|
| 4321 |
"addBosToken": true,
|
| 4322 |
"addEosToken": true
|
| 4323 |
+
},
|
| 4324 |
+
"artifactIdentity": {
|
| 4325 |
+
"sourceCheckpointId": "google/embeddinggemma-300m",
|
| 4326 |
+
"weightPackId": "google-embeddinggemma-300m-q4k-ehf16-af32-wp-catalog-v1",
|
| 4327 |
+
"manifestVariantId": "google-embeddinggemma-300m-q4k-ehf16-af32-mv-exec-v1",
|
| 4328 |
+
"artifactCompleteness": "complete"
|
| 4329 |
}
|
| 4330 |
+
}
|