clocksmith commited on
Commit
7e194b4
·
verified ·
1 Parent(s): b754c32

Publish gemma-3-270m-it-q4k-ehf16-af32 RDRR artifact

Browse files
models/gemma-3-270m-it-q4k-ehf16-af32/manifest.json CHANGED
@@ -141,29 +141,33 @@
141
  "maxBytes": 134217728,
142
  "outputDtype": "f32"
143
  }
144
- }
 
 
 
 
145
  },
146
  "execution": {
147
  "kernels": {
148
  "embed": {
149
  "kernel": "gather.wgsl",
150
  "entry": "main",
151
- "digest": "sha256:777991fb6e4b3b506e4493b47ee998afe541924ddd7c04e1eadf4cb7fd719ef8"
152
  },
153
  "rmsnorm": {
154
  "kernel": "rmsnorm.wgsl",
155
  "entry": "main",
156
- "digest": "sha256:f516b3e4bde2015f2a207c3ca5b8c9820c7809fa8f8d0786f90c568e0f1ac077"
157
  },
158
  "gemv": {
159
  "kernel": "matmul_gemv_subgroup.wgsl",
160
  "entry": "main_vec4",
161
- "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
162
  },
163
  "rope": {
164
  "kernel": "rope.wgsl",
165
  "entry": "main",
166
- "digest": "sha256:238e989aad7da731046cd54d7bb3c80a6f26daf5d7a8e435607161de902e5882"
167
  },
168
  "attn_decode": {
169
  "kernel": "attention_decode_online_f16kv.wgsl",
@@ -194,7 +198,7 @@
194
  "attn_small": {
195
  "kernel": "attention_small_f16kv.wgsl",
196
  "entry": "main",
197
- "digest": "sha256:7e3acc24b8b45294d18052c759319fe8a202d4ebfc4b653d4b904b245ae7e5c9",
198
  "precision": {
199
  "kvDtype": "f16"
200
  }
@@ -202,7 +206,7 @@
202
  "lm_head_gemv": {
203
  "kernel": "matmul_gemv_subgroup.wgsl",
204
  "entry": "main_multicol",
205
- "digest": "sha256:96c38c15e6fed0d7efdc5cd094db5843a8e8ddfe01eee3bc7322fa555dacf3d0",
206
  "constants": {
207
  "MULTICOL_COLS_PER_WG": 64,
208
  "MULTICOL_THREADS_PER_COL": 4
@@ -211,7 +215,7 @@
211
  "sample": {
212
  "kernel": "sample.wgsl",
213
  "entry": "sample_single_pass",
214
- "digest": "sha256:4412357e84113ee2f1bc0dc8bf89e314c2ab482c89c14ca016ea9949d16a9d0c"
215
  }
216
  },
217
  "preLayer": [
@@ -384,6 +388,9 @@
384
  "dtypeTransition": "require_cast_step",
385
  "unresolvedKernel": "error"
386
  }
 
 
 
387
  }
388
  },
389
  "shards": [
@@ -3333,5 +3340,11 @@
3333
  "vocabSize": 262144,
3334
  "file": "tokenizer.json",
3335
  "addBosToken": true
 
 
 
 
 
 
3336
  }
3337
- }
 
141
  "maxBytes": 134217728,
142
  "outputDtype": "f32"
143
  }
144
+ },
145
+ "prefillChunkSubmitMode": "sync",
146
+ "useFlashPrefillAttention": false,
147
+ "useWideTileQ4KPrefill": false,
148
+ "retainQ4KMaterialization": false
149
  },
150
  "execution": {
151
  "kernels": {
152
  "embed": {
153
  "kernel": "gather.wgsl",
154
  "entry": "main",
155
+ "digest": "sha256:4b12653c53247b32ebde7f6cf6a989d6248977e3816c761540b990b5f9818cb6"
156
  },
157
  "rmsnorm": {
158
  "kernel": "rmsnorm.wgsl",
159
  "entry": "main",
160
+ "digest": "sha256:c723c0300cc8f93396d79d1a274c0a90ccf1eb8ce82c0d778b8cd4659615fe45"
161
  },
162
  "gemv": {
163
  "kernel": "matmul_gemv_subgroup.wgsl",
164
  "entry": "main_vec4",
165
+ "digest": "sha256:2ea9aea659367210fbf61d922c0d4ee17cdac68ffb506b3bc5516e498071e011"
166
  },
167
  "rope": {
168
  "kernel": "rope.wgsl",
169
  "entry": "main",
170
+ "digest": "sha256:e13c54f261d7ef00474b71e4fe7b3298fb5cc49161f50c7ea1854aa6672ebe71"
171
  },
172
  "attn_decode": {
173
  "kernel": "attention_decode_online_f16kv.wgsl",
 
198
  "attn_small": {
199
  "kernel": "attention_small_f16kv.wgsl",
200
  "entry": "main",
201
+ "digest": "sha256:369aa3b386b4ef89a916ba2717465caa47079f52bbe0f3e5d61bd3220e2ec71e",
202
  "precision": {
203
  "kvDtype": "f16"
204
  }
 
206
  "lm_head_gemv": {
207
  "kernel": "matmul_gemv_subgroup.wgsl",
208
  "entry": "main_multicol",
209
+ "digest": "sha256:e41f94574d5ac54dd2710036da5d5acc643a483b79b2d74b86825efdaaa7438f",
210
  "constants": {
211
  "MULTICOL_COLS_PER_WG": 64,
212
  "MULTICOL_THREADS_PER_COL": 4
 
215
  "sample": {
216
  "kernel": "sample.wgsl",
217
  "entry": "sample_single_pass",
218
+ "digest": "sha256:7ff3dd675b4737d920e0b3889e15c4fedde8410a353bbc6d4651db76ce69eb9b"
219
  }
220
  },
221
  "preLayer": [
 
388
  "dtypeTransition": "require_cast_step",
389
  "unresolvedKernel": "error"
390
  }
391
+ },
392
+ "largeWeights": {
393
+ "gpuResidentOverrides": null
394
  }
395
  },
396
  "shards": [
 
3340
  "vocabSize": 262144,
3341
  "file": "tokenizer.json",
3342
  "addBosToken": true
3343
+ },
3344
+ "artifactIdentity": {
3345
+ "sourceCheckpointId": "google/gemma-3-270m-it",
3346
+ "weightPackId": "gemma-3-270m-it-q4k-ehf16-af32-wp-catalog-v1",
3347
+ "manifestVariantId": "gemma-3-270m-it-q4k-ehf16-af32-mv-exec-v1",
3348
+ "artifactCompleteness": "complete"
3349
  }
3350
+ }