Clocksmith commited on
Commit
66328ec
·
verified ·
1 Parent(s): 0363a91

Add Gemma 3 270M RDRR shards

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ models/gemma-3-270m-it-wq4k-ef16/tokenizer.json filter=lfs diff=lfs merge=lfs -text
models/gemma-3-270m-it-wq4k-ef16/manifest.json ADDED
@@ -0,0 +1,3503 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": 1,
3
+ "modelId": "gemma-3-270m-it-wq4k-ef16-hf16-f32",
4
+ "modelType": "transformer",
5
+ "quantization": "Q4_K_M",
6
+ "quantizationInfo": {
7
+ "weights": "q4k",
8
+ "embeddings": "f16",
9
+ "compute": "f32",
10
+ "layout": "row",
11
+ "variantTag": "wq4k-ef16"
12
+ },
13
+ "architecture": {
14
+ "numLayers": 18,
15
+ "hiddenSize": 640,
16
+ "intermediateSize": 2048,
17
+ "numAttentionHeads": 4,
18
+ "numKeyValueHeads": 1,
19
+ "headDim": 256,
20
+ "vocabSize": 262144,
21
+ "maxSeqLen": 32768,
22
+ "ropeTheta": 1000000
23
+ },
24
+ "moeConfig": null,
25
+ "inference": {
26
+ "schema": "doppler.execution/v0",
27
+ "presetId": "gemma3",
28
+ "attention": {
29
+ "queryPreAttnScalar": 16,
30
+ "attnLogitSoftcapping": null,
31
+ "slidingWindow": null,
32
+ "queryKeyNorm": true,
33
+ "causal": true,
34
+ "attentionBias": false
35
+ },
36
+ "normalization": {
37
+ "rmsNormEps": 0.000001,
38
+ "rmsNormWeightOffset": true,
39
+ "postAttentionNorm": true,
40
+ "preFeedforwardNorm": true,
41
+ "postFeedforwardNorm": true
42
+ },
43
+ "ffn": {
44
+ "activation": "gelu",
45
+ "gatedActivation": true,
46
+ "swigluLimit": null
47
+ },
48
+ "rope": {
49
+ "ropeTheta": 1000000,
50
+ "ropeLocalTheta": 10000,
51
+ "ropeScalingType": null,
52
+ "ropeScalingFactor": 1,
53
+ "yarnBetaFast": null,
54
+ "yarnBetaSlow": null,
55
+ "yarnOriginalMaxPos": null,
56
+ "ropeLocalScalingType": null,
57
+ "ropeLocalScalingFactor": 1,
58
+ "ropeLocalYarnBetaFast": null,
59
+ "ropeLocalYarnBetaSlow": null,
60
+ "ropeLocalYarnOriginalMaxPos": null
61
+ },
62
+ "output": {
63
+ "finalLogitSoftcapping": null,
64
+ "tieWordEmbeddings": true,
65
+ "scaleEmbeddings": true,
66
+ "embeddingTranspose": false,
67
+ "embeddingVocabSize": null
68
+ },
69
+ "layerPattern": {
70
+ "type": "every_n",
71
+ "globalPattern": null,
72
+ "period": 6,
73
+ "offset": 5
74
+ },
75
+ "chatTemplate": {
76
+ "type": "gemma",
77
+ "enabled": true
78
+ },
79
+ "pipeline": null,
80
+ "sessionDefaults": {
81
+ "compute": {
82
+ "defaults": {
83
+ "activationDtype": "f32",
84
+ "mathDtype": "f32",
85
+ "accumDtype": "f32",
86
+ "outputDtype": "f32"
87
+ },
88
+ "kernelProfiles": [
89
+ {
90
+ "kernelRef": {
91
+ "id": "gather.main",
92
+ "version": "1.0.0",
93
+ "digest": "sha256:777991fb6e4b3b506e4493b47ee998afe541924ddd7c04e1eadf4cb7fd719ef8"
94
+ }
95
+ },
96
+ {
97
+ "kernelRef": {
98
+ "id": "rmsnorm.main",
99
+ "version": "1.0.0",
100
+ "digest": "sha256:c529986befb29a04b94d89744585923a7cef82baf4b2b0a243aa2431618622cc"
101
+ }
102
+ },
103
+ {
104
+ "kernelRef": {
105
+ "id": "gather.f16.main",
106
+ "version": "1.0.0",
107
+ "digest": "sha256:a4829f4067091c98ad6ebbc9b0744cdd5bbcd4fbf6092b2f7cc7f1098695860f"
108
+ }
109
+ },
110
+ {
111
+ "kernelRef": {
112
+ "id": "matmul.gemv.subgroup.main.vec4",
113
+ "version": "1.0.0",
114
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
115
+ }
116
+ },
117
+ {
118
+ "kernelRef": {
119
+ "id": "rope.main",
120
+ "version": "1.0.0",
121
+ "digest": "sha256:b639fe8a54508115c82c13c923bfea89f59c6e15a5bef66bfc34e12f0ab4e32f"
122
+ }
123
+ },
124
+ {
125
+ "kernelRef": {
126
+ "id": "attention.decode.online.f16kv.main",
127
+ "version": "1.0.0",
128
+ "digest": "sha256:4c5d8c92a0a111af716d6b46b9559446807c086027445c7fefa150202f43dae4"
129
+ }
130
+ },
131
+ {
132
+ "kernelRef": {
133
+ "id": "residual.main",
134
+ "version": "1.0.0",
135
+ "digest": "sha256:1fc456b14e2fb2bc9627107b4e51e7a2098f723b5ba6ab5542cd9455af99f423"
136
+ }
137
+ },
138
+ {
139
+ "kernelRef": {
140
+ "id": "gelu.main",
141
+ "version": "1.0.0",
142
+ "digest": "sha256:a9007ea08aaff98f9be08f1e0490a6bcf252883eac5513de876ab9ce918865e6"
143
+ }
144
+ },
145
+ {
146
+ "kernelRef": {
147
+ "id": "matmul.f16w.f32a.tiled.main",
148
+ "version": "1.0.0",
149
+ "digest": "sha256:e94ae5374e8b43dd48b663eff59a45c822c3784d5702a1145266b6ffd15ba78c"
150
+ }
151
+ },
152
+ {
153
+ "kernelRef": {
154
+ "id": "matmul.f16w.f32a.main",
155
+ "version": "1.0.0",
156
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
157
+ }
158
+ },
159
+ {
160
+ "kernelRef": {
161
+ "id": "attention.small.f16kv.main",
162
+ "version": "1.0.0",
163
+ "digest": "sha256:7e3acc24b8b45294d18052c759319fe8a202d4ebfc4b653d4b904b245ae7e5c9"
164
+ }
165
+ },
166
+ {
167
+ "kernelRef": {
168
+ "id": "attention.streaming.f16kv.main",
169
+ "version": "1.0.0",
170
+ "digest": "sha256:b337a2dcf7b40431a733d1726eee8bf23504136fd5f915bec057fb59f3ea7480"
171
+ }
172
+ },
173
+ {
174
+ "kernelRef": {
175
+ "id": "matmul.gemv.subgroup.main.multicol",
176
+ "version": "1.0.0",
177
+ "digest": "sha256:96c38c15e6fed0d7efdc5cd094db5843a8e8ddfe01eee3bc7322fa555dacf3d0"
178
+ }
179
+ },
180
+ {
181
+ "kernelRef": {
182
+ "id": "sample.sample.single.pass",
183
+ "version": "1.0.0",
184
+ "digest": "sha256:4412357e84113ee2f1bc0dc8bf89e314c2ab482c89c14ca016ea9949d16a9d0c"
185
+ }
186
+ }
187
+ ]
188
+ },
189
+ "kvcache": {
190
+ "kvDtype": "f16"
191
+ },
192
+ "decodeLoop": null
193
+ },
194
+ "execution": {
195
+ "steps": [
196
+ {
197
+ "id": "preLayer_both_0_embed",
198
+ "phase": "both",
199
+ "section": "preLayer",
200
+ "op": "embed",
201
+ "kernel": "gather_f16.wgsl",
202
+ "entry": "main",
203
+ "weights": "embed_tokens",
204
+ "layers": "all",
205
+ "src": "state",
206
+ "dst": "state",
207
+ "kernelRef": {
208
+ "id": "gather.f16.main",
209
+ "version": "1.0.0",
210
+ "digest": "sha256:a4829f4067091c98ad6ebbc9b0744cdd5bbcd4fbf6092b2f7cc7f1098695860f"
211
+ }
212
+ },
213
+ {
214
+ "id": "layer_decode_1_input_norm",
215
+ "phase": "decode",
216
+ "section": "layer",
217
+ "op": "input_norm",
218
+ "kernel": "rmsnorm.wgsl",
219
+ "entry": "main",
220
+ "layers": "all",
221
+ "src": "state",
222
+ "dst": "state",
223
+ "kernelRef": {
224
+ "id": "rmsnorm.main",
225
+ "version": "1.0.0",
226
+ "digest": "sha256:c529986befb29a04b94d89744585923a7cef82baf4b2b0a243aa2431618622cc"
227
+ }
228
+ },
229
+ {
230
+ "id": "layer_decode_2_q_proj",
231
+ "phase": "decode",
232
+ "section": "layer",
233
+ "op": "q_proj",
234
+ "kernel": "matmul_gemv_subgroup.wgsl",
235
+ "entry": "main_vec4",
236
+ "weights": "layer.{L}.self_attn.q_proj",
237
+ "layers": "all",
238
+ "src": "state",
239
+ "dst": "state",
240
+ "kernelRef": {
241
+ "id": "matmul.gemv.subgroup.main.vec4",
242
+ "version": "1.0.0",
243
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
244
+ }
245
+ },
246
+ {
247
+ "id": "layer_decode_3_k_proj",
248
+ "phase": "decode",
249
+ "section": "layer",
250
+ "op": "k_proj",
251
+ "kernel": "matmul_gemv_subgroup.wgsl",
252
+ "entry": "main_vec4",
253
+ "weights": "layer.{L}.self_attn.k_proj",
254
+ "layers": "all",
255
+ "src": "state",
256
+ "dst": "state",
257
+ "kernelRef": {
258
+ "id": "matmul.gemv.subgroup.main.vec4",
259
+ "version": "1.0.0",
260
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
261
+ }
262
+ },
263
+ {
264
+ "id": "layer_decode_4_v_proj",
265
+ "phase": "decode",
266
+ "section": "layer",
267
+ "op": "v_proj",
268
+ "kernel": "matmul_gemv_subgroup.wgsl",
269
+ "entry": "main_vec4",
270
+ "weights": "layer.{L}.self_attn.v_proj",
271
+ "layers": "all",
272
+ "src": "state",
273
+ "dst": "state",
274
+ "kernelRef": {
275
+ "id": "matmul.gemv.subgroup.main.vec4",
276
+ "version": "1.0.0",
277
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
278
+ }
279
+ },
280
+ {
281
+ "id": "layer_decode_5_rope_q",
282
+ "phase": "decode",
283
+ "section": "layer",
284
+ "op": "rope_q",
285
+ "kernel": "rope.wgsl",
286
+ "entry": "main",
287
+ "layers": "all",
288
+ "src": "state",
289
+ "dst": "state",
290
+ "kernelRef": {
291
+ "id": "rope.main",
292
+ "version": "1.0.0",
293
+ "digest": "sha256:b639fe8a54508115c82c13c923bfea89f59c6e15a5bef66bfc34e12f0ab4e32f"
294
+ }
295
+ },
296
+ {
297
+ "id": "layer_decode_6_rope_k",
298
+ "phase": "decode",
299
+ "section": "layer",
300
+ "op": "rope_k",
301
+ "kernel": "rope.wgsl",
302
+ "entry": "main",
303
+ "layers": "all",
304
+ "src": "state",
305
+ "dst": "state",
306
+ "kernelRef": {
307
+ "id": "rope.main",
308
+ "version": "1.0.0",
309
+ "digest": "sha256:b639fe8a54508115c82c13c923bfea89f59c6e15a5bef66bfc34e12f0ab4e32f"
310
+ }
311
+ },
312
+ {
313
+ "id": "layer_decode_7_attention",
314
+ "phase": "decode",
315
+ "section": "layer",
316
+ "op": "attention",
317
+ "kernel": "attention_decode_online_f16kv.wgsl",
318
+ "entry": "main",
319
+ "layers": "all",
320
+ "src": "state",
321
+ "dst": "state",
322
+ "kernelRef": {
323
+ "id": "attention.decode.online.f16kv.main",
324
+ "version": "1.0.0",
325
+ "digest": "sha256:4c5d8c92a0a111af716d6b46b9559446807c086027445c7fefa150202f43dae4"
326
+ }
327
+ },
328
+ {
329
+ "id": "layer_decode_8_o_proj",
330
+ "phase": "decode",
331
+ "section": "layer",
332
+ "op": "o_proj",
333
+ "kernel": "matmul_gemv_subgroup.wgsl",
334
+ "entry": "main_vec4",
335
+ "weights": "layer.{L}.self_attn.o_proj",
336
+ "layers": "all",
337
+ "src": "state",
338
+ "dst": "state",
339
+ "kernelRef": {
340
+ "id": "matmul.gemv.subgroup.main.vec4",
341
+ "version": "1.0.0",
342
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
343
+ }
344
+ },
345
+ {
346
+ "id": "layer_decode_9_attn_residual",
347
+ "phase": "decode",
348
+ "section": "layer",
349
+ "op": "attn_residual",
350
+ "kernel": "residual.wgsl",
351
+ "entry": "main",
352
+ "layers": "all",
353
+ "src": "state",
354
+ "dst": "state",
355
+ "kernelRef": {
356
+ "id": "residual.main",
357
+ "version": "1.0.0",
358
+ "digest": "sha256:1fc456b14e2fb2bc9627107b4e51e7a2098f723b5ba6ab5542cd9455af99f423"
359
+ }
360
+ },
361
+ {
362
+ "id": "layer_decode_10_post_attn_norm",
363
+ "phase": "decode",
364
+ "section": "layer",
365
+ "op": "post_attn_norm",
366
+ "kernel": "rmsnorm.wgsl",
367
+ "entry": "main",
368
+ "layers": "all",
369
+ "src": "state",
370
+ "dst": "state",
371
+ "kernelRef": {
372
+ "id": "rmsnorm.main",
373
+ "version": "1.0.0",
374
+ "digest": "sha256:c529986befb29a04b94d89744585923a7cef82baf4b2b0a243aa2431618622cc"
375
+ }
376
+ },
377
+ {
378
+ "id": "layer_decode_11_gate_proj",
379
+ "phase": "decode",
380
+ "section": "layer",
381
+ "op": "gate_proj",
382
+ "kernel": "matmul_gemv_subgroup.wgsl",
383
+ "entry": "main_vec4",
384
+ "weights": "layer.{L}.mlp.gate_proj",
385
+ "layers": "all",
386
+ "src": "state",
387
+ "dst": "state",
388
+ "kernelRef": {
389
+ "id": "matmul.gemv.subgroup.main.vec4",
390
+ "version": "1.0.0",
391
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
392
+ }
393
+ },
394
+ {
395
+ "id": "layer_decode_12_up_proj",
396
+ "phase": "decode",
397
+ "section": "layer",
398
+ "op": "up_proj",
399
+ "kernel": "matmul_gemv_subgroup.wgsl",
400
+ "entry": "main_vec4",
401
+ "weights": "layer.{L}.mlp.up_proj",
402
+ "layers": "all",
403
+ "src": "state",
404
+ "dst": "state",
405
+ "kernelRef": {
406
+ "id": "matmul.gemv.subgroup.main.vec4",
407
+ "version": "1.0.0",
408
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
409
+ }
410
+ },
411
+ {
412
+ "id": "layer_decode_13_activation",
413
+ "phase": "decode",
414
+ "section": "layer",
415
+ "op": "activation",
416
+ "kernel": "gelu.wgsl",
417
+ "entry": "main",
418
+ "constants": {
419
+ "HAS_GATE": true
420
+ },
421
+ "layers": "all",
422
+ "src": "state",
423
+ "dst": "state",
424
+ "kernelRef": {
425
+ "id": "gelu.main",
426
+ "version": "1.0.0",
427
+ "digest": "sha256:a9007ea08aaff98f9be08f1e0490a6bcf252883eac5513de876ab9ce918865e6"
428
+ }
429
+ },
430
+ {
431
+ "id": "layer_decode_14_down_proj",
432
+ "phase": "decode",
433
+ "section": "layer",
434
+ "op": "down_proj",
435
+ "kernel": "matmul_gemv_subgroup.wgsl",
436
+ "entry": "main_vec4",
437
+ "weights": "layer.{L}.mlp.down_proj",
438
+ "layers": "all",
439
+ "src": "state",
440
+ "dst": "state",
441
+ "kernelRef": {
442
+ "id": "matmul.gemv.subgroup.main.vec4",
443
+ "version": "1.0.0",
444
+ "digest": "sha256:3cee3bed453b40c5564a751d2a917649e10ad52f5268e77cbfecfcee34780457"
445
+ }
446
+ },
447
+ {
448
+ "id": "layer_decode_15_ffn_residual",
449
+ "phase": "decode",
450
+ "section": "layer",
451
+ "op": "ffn_residual",
452
+ "kernel": "residual.wgsl",
453
+ "entry": "main",
454
+ "layers": "all",
455
+ "src": "state",
456
+ "dst": "state",
457
+ "kernelRef": {
458
+ "id": "residual.main",
459
+ "version": "1.0.0",
460
+ "digest": "sha256:1fc456b14e2fb2bc9627107b4e51e7a2098f723b5ba6ab5542cd9455af99f423"
461
+ }
462
+ },
463
+ {
464
+ "id": "layer_prefill_16_input_norm",
465
+ "phase": "prefill",
466
+ "section": "layer",
467
+ "op": "input_norm",
468
+ "kernel": "rmsnorm.wgsl",
469
+ "entry": "main",
470
+ "layers": "all",
471
+ "src": "state",
472
+ "dst": "state",
473
+ "kernelRef": {
474
+ "id": "rmsnorm.main",
475
+ "version": "1.0.0",
476
+ "digest": "sha256:c529986befb29a04b94d89744585923a7cef82baf4b2b0a243aa2431618622cc"
477
+ }
478
+ },
479
+ {
480
+ "id": "layer_prefill_17_q_proj",
481
+ "phase": "prefill",
482
+ "section": "layer",
483
+ "op": "q_proj",
484
+ "kernel": "matmul_f16w_f32a.wgsl",
485
+ "entry": "main",
486
+ "weights": "layer.{L}.self_attn.q_proj",
487
+ "layers": "all",
488
+ "src": "state",
489
+ "dst": "state",
490
+ "kernelRef": {
491
+ "id": "matmul.f16w.f32a.main",
492
+ "version": "1.0.0",
493
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
494
+ }
495
+ },
496
+ {
497
+ "id": "layer_prefill_18_k_proj",
498
+ "phase": "prefill",
499
+ "section": "layer",
500
+ "op": "k_proj",
501
+ "kernel": "matmul_f16w_f32a.wgsl",
502
+ "entry": "main",
503
+ "weights": "layer.{L}.self_attn.k_proj",
504
+ "layers": "all",
505
+ "src": "state",
506
+ "dst": "state",
507
+ "kernelRef": {
508
+ "id": "matmul.f16w.f32a.main",
509
+ "version": "1.0.0",
510
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
511
+ }
512
+ },
513
+ {
514
+ "id": "layer_prefill_19_v_proj",
515
+ "phase": "prefill",
516
+ "section": "layer",
517
+ "op": "v_proj",
518
+ "kernel": "matmul_f16w_f32a.wgsl",
519
+ "entry": "main",
520
+ "weights": "layer.{L}.self_attn.v_proj",
521
+ "layers": "all",
522
+ "src": "state",
523
+ "dst": "state",
524
+ "kernelRef": {
525
+ "id": "matmul.f16w.f32a.main",
526
+ "version": "1.0.0",
527
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
528
+ }
529
+ },
530
+ {
531
+ "id": "layer_prefill_20_rope_q",
532
+ "phase": "prefill",
533
+ "section": "layer",
534
+ "op": "rope_q",
535
+ "kernel": "rope.wgsl",
536
+ "entry": "main",
537
+ "layers": "all",
538
+ "src": "state",
539
+ "dst": "state",
540
+ "kernelRef": {
541
+ "id": "rope.main",
542
+ "version": "1.0.0",
543
+ "digest": "sha256:b639fe8a54508115c82c13c923bfea89f59c6e15a5bef66bfc34e12f0ab4e32f"
544
+ }
545
+ },
546
+ {
547
+ "id": "layer_prefill_21_rope_k",
548
+ "phase": "prefill",
549
+ "section": "layer",
550
+ "op": "rope_k",
551
+ "kernel": "rope.wgsl",
552
+ "entry": "main",
553
+ "layers": "all",
554
+ "src": "state",
555
+ "dst": "state",
556
+ "kernelRef": {
557
+ "id": "rope.main",
558
+ "version": "1.0.0",
559
+ "digest": "sha256:b639fe8a54508115c82c13c923bfea89f59c6e15a5bef66bfc34e12f0ab4e32f"
560
+ }
561
+ },
562
+ {
563
+ "id": "layer_prefill_22_attention",
564
+ "phase": "prefill",
565
+ "section": "layer",
566
+ "op": "attention",
567
+ "kernel": "attention_streaming_f16kv.wgsl",
568
+ "entry": "main",
569
+ "layers": "all",
570
+ "src": "state",
571
+ "dst": "state",
572
+ "kernelRef": {
573
+ "id": "attention.streaming.f16kv.main",
574
+ "version": "1.0.0",
575
+ "digest": "sha256:b337a2dcf7b40431a733d1726eee8bf23504136fd5f915bec057fb59f3ea7480"
576
+ }
577
+ },
578
+ {
579
+ "id": "layer_prefill_23_o_proj",
580
+ "phase": "prefill",
581
+ "section": "layer",
582
+ "op": "o_proj",
583
+ "kernel": "matmul_f16w_f32a.wgsl",
584
+ "entry": "main",
585
+ "weights": "layer.{L}.self_attn.o_proj",
586
+ "layers": "all",
587
+ "src": "state",
588
+ "dst": "state",
589
+ "kernelRef": {
590
+ "id": "matmul.f16w.f32a.main",
591
+ "version": "1.0.0",
592
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
593
+ }
594
+ },
595
+ {
596
+ "id": "layer_prefill_24_attn_residual",
597
+ "phase": "prefill",
598
+ "section": "layer",
599
+ "op": "attn_residual",
600
+ "kernel": "residual.wgsl",
601
+ "entry": "main",
602
+ "layers": "all",
603
+ "src": "state",
604
+ "dst": "state",
605
+ "kernelRef": {
606
+ "id": "residual.main",
607
+ "version": "1.0.0",
608
+ "digest": "sha256:1fc456b14e2fb2bc9627107b4e51e7a2098f723b5ba6ab5542cd9455af99f423"
609
+ }
610
+ },
611
+ {
612
+ "id": "layer_prefill_25_post_attn_norm",
613
+ "phase": "prefill",
614
+ "section": "layer",
615
+ "op": "post_attn_norm",
616
+ "kernel": "rmsnorm.wgsl",
617
+ "entry": "main",
618
+ "layers": "all",
619
+ "src": "state",
620
+ "dst": "state",
621
+ "kernelRef": {
622
+ "id": "rmsnorm.main",
623
+ "version": "1.0.0",
624
+ "digest": "sha256:c529986befb29a04b94d89744585923a7cef82baf4b2b0a243aa2431618622cc"
625
+ }
626
+ },
627
+ {
628
+ "id": "layer_prefill_26_gate_proj",
629
+ "phase": "prefill",
630
+ "section": "layer",
631
+ "op": "gate_proj",
632
+ "kernel": "matmul_f16w_f32a.wgsl",
633
+ "entry": "main",
634
+ "weights": "layer.{L}.mlp.gate_proj",
635
+ "layers": "all",
636
+ "src": "state",
637
+ "dst": "state",
638
+ "kernelRef": {
639
+ "id": "matmul.f16w.f32a.main",
640
+ "version": "1.0.0",
641
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
642
+ }
643
+ },
644
+ {
645
+ "id": "layer_prefill_27_up_proj",
646
+ "phase": "prefill",
647
+ "section": "layer",
648
+ "op": "up_proj",
649
+ "kernel": "matmul_f16w_f32a.wgsl",
650
+ "entry": "main",
651
+ "weights": "layer.{L}.mlp.up_proj",
652
+ "layers": "all",
653
+ "src": "state",
654
+ "dst": "state",
655
+ "kernelRef": {
656
+ "id": "matmul.f16w.f32a.main",
657
+ "version": "1.0.0",
658
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
659
+ }
660
+ },
661
+ {
662
+ "id": "layer_prefill_28_activation",
663
+ "phase": "prefill",
664
+ "section": "layer",
665
+ "op": "activation",
666
+ "kernel": "gelu.wgsl",
667
+ "entry": "main",
668
+ "constants": {
669
+ "HAS_GATE": true
670
+ },
671
+ "layers": "all",
672
+ "src": "state",
673
+ "dst": "state",
674
+ "kernelRef": {
675
+ "id": "gelu.main",
676
+ "version": "1.0.0",
677
+ "digest": "sha256:a9007ea08aaff98f9be08f1e0490a6bcf252883eac5513de876ab9ce918865e6"
678
+ }
679
+ },
680
+ {
681
+ "id": "layer_prefill_29_down_proj",
682
+ "phase": "prefill",
683
+ "section": "layer",
684
+ "op": "down_proj",
685
+ "kernel": "matmul_f16w_f32a.wgsl",
686
+ "entry": "main",
687
+ "weights": "layer.{L}.mlp.down_proj",
688
+ "layers": "all",
689
+ "src": "state",
690
+ "dst": "state",
691
+ "kernelRef": {
692
+ "id": "matmul.f16w.f32a.main",
693
+ "version": "1.0.0",
694
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
695
+ }
696
+ },
697
+ {
698
+ "id": "layer_prefill_30_ffn_residual",
699
+ "phase": "prefill",
700
+ "section": "layer",
701
+ "op": "ffn_residual",
702
+ "kernel": "residual.wgsl",
703
+ "entry": "main",
704
+ "layers": "all",
705
+ "src": "state",
706
+ "dst": "state",
707
+ "kernelRef": {
708
+ "id": "residual.main",
709
+ "version": "1.0.0",
710
+ "digest": "sha256:1fc456b14e2fb2bc9627107b4e51e7a2098f723b5ba6ab5542cd9455af99f423"
711
+ }
712
+ },
713
+ {
714
+ "id": "postLayer_both_31_final_norm",
715
+ "phase": "both",
716
+ "section": "postLayer",
717
+ "op": "final_norm",
718
+ "kernel": "rmsnorm.wgsl",
719
+ "entry": "main",
720
+ "layers": "all",
721
+ "src": "state",
722
+ "dst": "state",
723
+ "kernelRef": {
724
+ "id": "rmsnorm.main",
725
+ "version": "1.0.0",
726
+ "digest": "sha256:c529986befb29a04b94d89744585923a7cef82baf4b2b0a243aa2431618622cc"
727
+ }
728
+ },
729
+ {
730
+ "id": "postLayer_both_32_lm_head",
731
+ "phase": "both",
732
+ "section": "postLayer",
733
+ "op": "lm_head",
734
+ "kernel": "matmul_gemv_subgroup.wgsl",
735
+ "entry": "main_multicol",
736
+ "weights": "lm_head",
737
+ "constants": {
738
+ "MULTICOL_COLS_PER_WG": 64,
739
+ "MULTICOL_THREADS_PER_COL": 4
740
+ },
741
+ "layers": "all",
742
+ "src": "state",
743
+ "dst": "state",
744
+ "kernelRef": {
745
+ "id": "matmul.gemv.subgroup.main.multicol",
746
+ "version": "1.0.0",
747
+ "digest": "sha256:96c38c15e6fed0d7efdc5cd094db5843a8e8ddfe01eee3bc7322fa555dacf3d0"
748
+ }
749
+ },
750
+ {
751
+ "id": "postLayer_both_33_lm_head_prefill",
752
+ "phase": "both",
753
+ "section": "postLayer",
754
+ "op": "lm_head_prefill",
755
+ "kernel": "matmul_f16w_f32a.wgsl",
756
+ "entry": "main",
757
+ "weights": "lm_head",
758
+ "layers": "all",
759
+ "src": "state",
760
+ "dst": "state",
761
+ "kernelRef": {
762
+ "id": "matmul.f16w.f32a.main",
763
+ "version": "1.0.0",
764
+ "digest": "sha256:027a8f1cd9713cbe0b0ada160bd175e0542bb90896ad85441b023522d9a1befc"
765
+ }
766
+ },
767
+ {
768
+ "id": "sampling_decode_34_sample",
769
+ "phase": "decode",
770
+ "section": "sampling",
771
+ "op": "sample",
772
+ "kernel": "sample.wgsl",
773
+ "entry": "sample_single_pass",
774
+ "layers": "all",
775
+ "src": "state",
776
+ "dst": "state",
777
+ "kernelRef": {
778
+ "id": "sample.sample.single.pass",
779
+ "version": "1.0.0",
780
+ "digest": "sha256:4412357e84113ee2f1bc0dc8bf89e314c2ab482c89c14ca016ea9949d16a9d0c"
781
+ }
782
+ }
783
+ ],
784
+ "policies": {
785
+ "precisionPrecedence": "step_then_kernel_profile_then_session_default",
786
+ "unsupportedPrecision": "error",
787
+ "dtypeTransition": "require_cast_step",
788
+ "unresolvedKernel": "error"
789
+ }
790
+ },
791
+ "defaultKernelPath": "gemma3-q4k-dequant-f32a-online"
792
+ },
793
+ "shards": [
794
+ {
795
+ "index": 0,
796
+ "filename": "shard_00000.bin",
797
+ "size": 67108864,
798
+ "hash": "3f4369beb50460c76843aab54e36ab042fedffd0b0c03d09e39ec4fc995e6c67",
799
+ "blake3": "3f4369beb50460c76843aab54e36ab042fedffd0b0c03d09e39ec4fc995e6c67",
800
+ "offset": 0
801
+ },
802
+ {
803
+ "index": 1,
804
+ "filename": "shard_00001.bin",
805
+ "size": 67108864,
806
+ "hash": "7c9b6e74ef5738a687bbc28b80007ba154ad529ba121c3eecd3f73ac14b98e16",
807
+ "blake3": "7c9b6e74ef5738a687bbc28b80007ba154ad529ba121c3eecd3f73ac14b98e16",
808
+ "offset": 67108864
809
+ },
810
+ {
811
+ "index": 2,
812
+ "filename": "shard_00002.bin",
813
+ "size": 67108864,
814
+ "hash": "1e9affca30a11b90d2cb02ce15aaab4e29a0d031c7696bceed344e7b80f7870b",
815
+ "blake3": "1e9affca30a11b90d2cb02ce15aaab4e29a0d031c7696bceed344e7b80f7870b",
816
+ "offset": 134217728
817
+ },
818
+ {
819
+ "index": 3,
820
+ "filename": "shard_00003.bin",
821
+ "size": 67108864,
822
+ "hash": "bdc8196efc0b2f4442bfb3aeff5097b96172bc4527693bd0bb8df5856d9fc496",
823
+ "blake3": "bdc8196efc0b2f4442bfb3aeff5097b96172bc4527693bd0bb8df5856d9fc496",
824
+ "offset": 201326592
825
+ },
826
+ {
827
+ "index": 4,
828
+ "filename": "shard_00004.bin",
829
+ "size": 67108864,
830
+ "hash": "9e1391f8d5d66008df9a1fefbdf8c24105ac7f1c1b5340e474162e33a8480a30",
831
+ "blake3": "9e1391f8d5d66008df9a1fefbdf8c24105ac7f1c1b5340e474162e33a8480a30",
832
+ "offset": 268435456
833
+ },
834
+ {
835
+ "index": 5,
836
+ "filename": "shard_00005.bin",
837
+ "size": 63812864,
838
+ "hash": "b0bae3311c0c8b9edeec774e25f585f233b31ca79d37e5709eed733c7a932031",
839
+ "blake3": "b0bae3311c0c8b9edeec774e25f585f233b31ca79d37e5709eed733c7a932031",
840
+ "offset": 335544320
841
+ }
842
+ ],
843
+ "tensors": {
844
+ "model.embed_tokens.weight": {
845
+ "spans": [
846
+ {
847
+ "shardIndex": 0,
848
+ "offset": 0,
849
+ "size": 67108864
850
+ },
851
+ {
852
+ "shardIndex": 1,
853
+ "offset": 0,
854
+ "size": 67108864
855
+ },
856
+ {
857
+ "shardIndex": 2,
858
+ "offset": 0,
859
+ "size": 67108864
860
+ },
861
+ {
862
+ "shardIndex": 3,
863
+ "offset": 0,
864
+ "size": 67108864
865
+ },
866
+ {
867
+ "shardIndex": 4,
868
+ "offset": 0,
869
+ "size": 67108864
870
+ }
871
+ ],
872
+ "size": 335544320,
873
+ "shape": [
874
+ 262144,
875
+ 640
876
+ ],
877
+ "dtype": "F16",
878
+ "role": "embedding"
879
+ },
880
+ "model.layers.0.input_layernorm.weight": {
881
+ "shard": 5,
882
+ "offset": 0,
883
+ "size": 1280,
884
+ "shape": [
885
+ 640
886
+ ],
887
+ "dtype": "BF16",
888
+ "role": "norm"
889
+ },
890
+ "model.layers.0.mlp.down_proj.weight": {
891
+ "shard": 5,
892
+ "offset": 1280,
893
+ "size": 737280,
894
+ "shape": [
895
+ 640,
896
+ 2048
897
+ ],
898
+ "dtype": "Q4_K_M",
899
+ "role": "matmul",
900
+ "layout": "row"
901
+ },
902
+ "model.layers.0.mlp.gate_proj.weight": {
903
+ "shard": 5,
904
+ "offset": 738560,
905
+ "size": 884736,
906
+ "shape": [
907
+ 2048,
908
+ 640
909
+ ],
910
+ "dtype": "Q4_K_M",
911
+ "role": "matmul",
912
+ "layout": "row"
913
+ },
914
+ "model.layers.0.mlp.up_proj.weight": {
915
+ "shard": 5,
916
+ "offset": 1623296,
917
+ "size": 884736,
918
+ "shape": [
919
+ 2048,
920
+ 640
921
+ ],
922
+ "dtype": "Q4_K_M",
923
+ "role": "matmul",
924
+ "layout": "row"
925
+ },
926
+ "model.layers.0.post_attention_layernorm.weight": {
927
+ "shard": 5,
928
+ "offset": 2508032,
929
+ "size": 1280,
930
+ "shape": [
931
+ 640
932
+ ],
933
+ "dtype": "BF16",
934
+ "role": "norm"
935
+ },
936
+ "model.layers.0.post_feedforward_layernorm.weight": {
937
+ "shard": 5,
938
+ "offset": 2509312,
939
+ "size": 1280,
940
+ "shape": [
941
+ 640
942
+ ],
943
+ "dtype": "BF16",
944
+ "role": "norm"
945
+ },
946
+ "model.layers.0.pre_feedforward_layernorm.weight": {
947
+ "shard": 5,
948
+ "offset": 2510592,
949
+ "size": 1280,
950
+ "shape": [
951
+ 640
952
+ ],
953
+ "dtype": "BF16",
954
+ "role": "norm"
955
+ },
956
+ "model.layers.0.self_attn.k_norm.weight": {
957
+ "shard": 5,
958
+ "offset": 2511872,
959
+ "size": 512,
960
+ "shape": [
961
+ 256
962
+ ],
963
+ "dtype": "BF16",
964
+ "role": "norm"
965
+ },
966
+ "model.layers.0.self_attn.k_proj.weight": {
967
+ "shard": 5,
968
+ "offset": 2512384,
969
+ "size": 110592,
970
+ "shape": [
971
+ 256,
972
+ 640
973
+ ],
974
+ "dtype": "Q4_K_M",
975
+ "role": "matmul",
976
+ "layout": "row"
977
+ },
978
+ "model.layers.0.self_attn.o_proj.weight": {
979
+ "shard": 5,
980
+ "offset": 2622976,
981
+ "size": 368640,
982
+ "shape": [
983
+ 640,
984
+ 1024
985
+ ],
986
+ "dtype": "Q4_K_M",
987
+ "role": "matmul",
988
+ "layout": "row"
989
+ },
990
+ "model.layers.0.self_attn.q_norm.weight": {
991
+ "shard": 5,
992
+ "offset": 2991616,
993
+ "size": 512,
994
+ "shape": [
995
+ 256
996
+ ],
997
+ "dtype": "BF16",
998
+ "role": "norm"
999
+ },
1000
+ "model.layers.0.self_attn.q_proj.weight": {
1001
+ "shard": 5,
1002
+ "offset": 2992128,
1003
+ "size": 442368,
1004
+ "shape": [
1005
+ 1024,
1006
+ 640
1007
+ ],
1008
+ "dtype": "Q4_K_M",
1009
+ "role": "matmul",
1010
+ "layout": "row"
1011
+ },
1012
+ "model.layers.0.self_attn.v_proj.weight": {
1013
+ "shard": 5,
1014
+ "offset": 3434496,
1015
+ "size": 110592,
1016
+ "shape": [
1017
+ 256,
1018
+ 640
1019
+ ],
1020
+ "dtype": "Q4_K_M",
1021
+ "role": "matmul",
1022
+ "layout": "row"
1023
+ },
1024
+ "model.layers.1.input_layernorm.weight": {
1025
+ "shard": 5,
1026
+ "offset": 3545088,
1027
+ "size": 1280,
1028
+ "shape": [
1029
+ 640
1030
+ ],
1031
+ "dtype": "BF16",
1032
+ "role": "norm"
1033
+ },
1034
+ "model.layers.1.mlp.down_proj.weight": {
1035
+ "shard": 5,
1036
+ "offset": 3546368,
1037
+ "size": 737280,
1038
+ "shape": [
1039
+ 640,
1040
+ 2048
1041
+ ],
1042
+ "dtype": "Q4_K_M",
1043
+ "role": "matmul",
1044
+ "layout": "row"
1045
+ },
1046
+ "model.layers.1.mlp.gate_proj.weight": {
1047
+ "shard": 5,
1048
+ "offset": 4283648,
1049
+ "size": 884736,
1050
+ "shape": [
1051
+ 2048,
1052
+ 640
1053
+ ],
1054
+ "dtype": "Q4_K_M",
1055
+ "role": "matmul",
1056
+ "layout": "row"
1057
+ },
1058
+ "model.layers.1.mlp.up_proj.weight": {
1059
+ "shard": 5,
1060
+ "offset": 5168384,
1061
+ "size": 884736,
1062
+ "shape": [
1063
+ 2048,
1064
+ 640
1065
+ ],
1066
+ "dtype": "Q4_K_M",
1067
+ "role": "matmul",
1068
+ "layout": "row"
1069
+ },
1070
+ "model.layers.1.post_attention_layernorm.weight": {
1071
+ "shard": 5,
1072
+ "offset": 6053120,
1073
+ "size": 1280,
1074
+ "shape": [
1075
+ 640
1076
+ ],
1077
+ "dtype": "BF16",
1078
+ "role": "norm"
1079
+ },
1080
+ "model.layers.1.post_feedforward_layernorm.weight": {
1081
+ "shard": 5,
1082
+ "offset": 6054400,
1083
+ "size": 1280,
1084
+ "shape": [
1085
+ 640
1086
+ ],
1087
+ "dtype": "BF16",
1088
+ "role": "norm"
1089
+ },
1090
+ "model.layers.1.pre_feedforward_layernorm.weight": {
1091
+ "shard": 5,
1092
+ "offset": 6055680,
1093
+ "size": 1280,
1094
+ "shape": [
1095
+ 640
1096
+ ],
1097
+ "dtype": "BF16",
1098
+ "role": "norm"
1099
+ },
1100
+ "model.layers.1.self_attn.k_norm.weight": {
1101
+ "shard": 5,
1102
+ "offset": 6056960,
1103
+ "size": 512,
1104
+ "shape": [
1105
+ 256
1106
+ ],
1107
+ "dtype": "BF16",
1108
+ "role": "norm"
1109
+ },
1110
+ "model.layers.1.self_attn.k_proj.weight": {
1111
+ "shard": 5,
1112
+ "offset": 6057472,
1113
+ "size": 110592,
1114
+ "shape": [
1115
+ 256,
1116
+ 640
1117
+ ],
1118
+ "dtype": "Q4_K_M",
1119
+ "role": "matmul",
1120
+ "layout": "row"
1121
+ },
1122
+ "model.layers.1.self_attn.o_proj.weight": {
1123
+ "shard": 5,
1124
+ "offset": 6168064,
1125
+ "size": 368640,
1126
+ "shape": [
1127
+ 640,
1128
+ 1024
1129
+ ],
1130
+ "dtype": "Q4_K_M",
1131
+ "role": "matmul",
1132
+ "layout": "row"
1133
+ },
1134
+ "model.layers.1.self_attn.q_norm.weight": {
1135
+ "shard": 5,
1136
+ "offset": 6536704,
1137
+ "size": 512,
1138
+ "shape": [
1139
+ 256
1140
+ ],
1141
+ "dtype": "BF16",
1142
+ "role": "norm"
1143
+ },
1144
+ "model.layers.1.self_attn.q_proj.weight": {
1145
+ "shard": 5,
1146
+ "offset": 6537216,
1147
+ "size": 442368,
1148
+ "shape": [
1149
+ 1024,
1150
+ 640
1151
+ ],
1152
+ "dtype": "Q4_K_M",
1153
+ "role": "matmul",
1154
+ "layout": "row"
1155
+ },
1156
+ "model.layers.1.self_attn.v_proj.weight": {
1157
+ "shard": 5,
1158
+ "offset": 6979584,
1159
+ "size": 110592,
1160
+ "shape": [
1161
+ 256,
1162
+ 640
1163
+ ],
1164
+ "dtype": "Q4_K_M",
1165
+ "role": "matmul",
1166
+ "layout": "row"
1167
+ },
1168
+ "model.layers.10.input_layernorm.weight": {
1169
+ "shard": 5,
1170
+ "offset": 7090176,
1171
+ "size": 1280,
1172
+ "shape": [
1173
+ 640
1174
+ ],
1175
+ "dtype": "BF16",
1176
+ "role": "norm"
1177
+ },
1178
+ "model.layers.10.mlp.down_proj.weight": {
1179
+ "shard": 5,
1180
+ "offset": 7091456,
1181
+ "size": 737280,
1182
+ "shape": [
1183
+ 640,
1184
+ 2048
1185
+ ],
1186
+ "dtype": "Q4_K_M",
1187
+ "role": "matmul",
1188
+ "layout": "row"
1189
+ },
1190
+ "model.layers.10.mlp.gate_proj.weight": {
1191
+ "shard": 5,
1192
+ "offset": 7828736,
1193
+ "size": 884736,
1194
+ "shape": [
1195
+ 2048,
1196
+ 640
1197
+ ],
1198
+ "dtype": "Q4_K_M",
1199
+ "role": "matmul",
1200
+ "layout": "row"
1201
+ },
1202
+ "model.layers.10.mlp.up_proj.weight": {
1203
+ "shard": 5,
1204
+ "offset": 8713472,
1205
+ "size": 884736,
1206
+ "shape": [
1207
+ 2048,
1208
+ 640
1209
+ ],
1210
+ "dtype": "Q4_K_M",
1211
+ "role": "matmul",
1212
+ "layout": "row"
1213
+ },
1214
+ "model.layers.10.post_attention_layernorm.weight": {
1215
+ "shard": 5,
1216
+ "offset": 9598208,
1217
+ "size": 1280,
1218
+ "shape": [
1219
+ 640
1220
+ ],
1221
+ "dtype": "BF16",
1222
+ "role": "norm"
1223
+ },
1224
+ "model.layers.10.post_feedforward_layernorm.weight": {
1225
+ "shard": 5,
1226
+ "offset": 9599488,
1227
+ "size": 1280,
1228
+ "shape": [
1229
+ 640
1230
+ ],
1231
+ "dtype": "BF16",
1232
+ "role": "norm"
1233
+ },
1234
+ "model.layers.10.pre_feedforward_layernorm.weight": {
1235
+ "shard": 5,
1236
+ "offset": 9600768,
1237
+ "size": 1280,
1238
+ "shape": [
1239
+ 640
1240
+ ],
1241
+ "dtype": "BF16",
1242
+ "role": "norm"
1243
+ },
1244
+ "model.layers.10.self_attn.k_norm.weight": {
1245
+ "shard": 5,
1246
+ "offset": 9602048,
1247
+ "size": 512,
1248
+ "shape": [
1249
+ 256
1250
+ ],
1251
+ "dtype": "BF16",
1252
+ "role": "norm"
1253
+ },
1254
+ "model.layers.10.self_attn.k_proj.weight": {
1255
+ "shard": 5,
1256
+ "offset": 9602560,
1257
+ "size": 110592,
1258
+ "shape": [
1259
+ 256,
1260
+ 640
1261
+ ],
1262
+ "dtype": "Q4_K_M",
1263
+ "role": "matmul",
1264
+ "layout": "row"
1265
+ },
1266
+ "model.layers.10.self_attn.o_proj.weight": {
1267
+ "shard": 5,
1268
+ "offset": 9713152,
1269
+ "size": 368640,
1270
+ "shape": [
1271
+ 640,
1272
+ 1024
1273
+ ],
1274
+ "dtype": "Q4_K_M",
1275
+ "role": "matmul",
1276
+ "layout": "row"
1277
+ },
1278
+ "model.layers.10.self_attn.q_norm.weight": {
1279
+ "shard": 5,
1280
+ "offset": 10081792,
1281
+ "size": 512,
1282
+ "shape": [
1283
+ 256
1284
+ ],
1285
+ "dtype": "BF16",
1286
+ "role": "norm"
1287
+ },
1288
+ "model.layers.10.self_attn.q_proj.weight": {
1289
+ "shard": 5,
1290
+ "offset": 10082304,
1291
+ "size": 442368,
1292
+ "shape": [
1293
+ 1024,
1294
+ 640
1295
+ ],
1296
+ "dtype": "Q4_K_M",
1297
+ "role": "matmul",
1298
+ "layout": "row"
1299
+ },
1300
+ "model.layers.10.self_attn.v_proj.weight": {
1301
+ "shard": 5,
1302
+ "offset": 10524672,
1303
+ "size": 110592,
1304
+ "shape": [
1305
+ 256,
1306
+ 640
1307
+ ],
1308
+ "dtype": "Q4_K_M",
1309
+ "role": "matmul",
1310
+ "layout": "row"
1311
+ },
1312
+ "model.layers.11.input_layernorm.weight": {
1313
+ "shard": 5,
1314
+ "offset": 10635264,
1315
+ "size": 1280,
1316
+ "shape": [
1317
+ 640
1318
+ ],
1319
+ "dtype": "BF16",
1320
+ "role": "norm"
1321
+ },
1322
+ "model.layers.11.mlp.down_proj.weight": {
1323
+ "shard": 5,
1324
+ "offset": 10636544,
1325
+ "size": 737280,
1326
+ "shape": [
1327
+ 640,
1328
+ 2048
1329
+ ],
1330
+ "dtype": "Q4_K_M",
1331
+ "role": "matmul",
1332
+ "layout": "row"
1333
+ },
1334
+ "model.layers.11.mlp.gate_proj.weight": {
1335
+ "shard": 5,
1336
+ "offset": 11373824,
1337
+ "size": 884736,
1338
+ "shape": [
1339
+ 2048,
1340
+ 640
1341
+ ],
1342
+ "dtype": "Q4_K_M",
1343
+ "role": "matmul",
1344
+ "layout": "row"
1345
+ },
1346
+ "model.layers.11.mlp.up_proj.weight": {
1347
+ "shard": 5,
1348
+ "offset": 12258560,
1349
+ "size": 884736,
1350
+ "shape": [
1351
+ 2048,
1352
+ 640
1353
+ ],
1354
+ "dtype": "Q4_K_M",
1355
+ "role": "matmul",
1356
+ "layout": "row"
1357
+ },
1358
+ "model.layers.11.post_attention_layernorm.weight": {
1359
+ "shard": 5,
1360
+ "offset": 13143296,
1361
+ "size": 1280,
1362
+ "shape": [
1363
+ 640
1364
+ ],
1365
+ "dtype": "BF16",
1366
+ "role": "norm"
1367
+ },
1368
+ "model.layers.11.post_feedforward_layernorm.weight": {
1369
+ "shard": 5,
1370
+ "offset": 13144576,
1371
+ "size": 1280,
1372
+ "shape": [
1373
+ 640
1374
+ ],
1375
+ "dtype": "BF16",
1376
+ "role": "norm"
1377
+ },
1378
+ "model.layers.11.pre_feedforward_layernorm.weight": {
1379
+ "shard": 5,
1380
+ "offset": 13145856,
1381
+ "size": 1280,
1382
+ "shape": [
1383
+ 640
1384
+ ],
1385
+ "dtype": "BF16",
1386
+ "role": "norm"
1387
+ },
1388
+ "model.layers.11.self_attn.k_norm.weight": {
1389
+ "shard": 5,
1390
+ "offset": 13147136,
1391
+ "size": 512,
1392
+ "shape": [
1393
+ 256
1394
+ ],
1395
+ "dtype": "BF16",
1396
+ "role": "norm"
1397
+ },
1398
+ "model.layers.11.self_attn.k_proj.weight": {
1399
+ "shard": 5,
1400
+ "offset": 13147648,
1401
+ "size": 110592,
1402
+ "shape": [
1403
+ 256,
1404
+ 640
1405
+ ],
1406
+ "dtype": "Q4_K_M",
1407
+ "role": "matmul",
1408
+ "layout": "row"
1409
+ },
1410
+ "model.layers.11.self_attn.o_proj.weight": {
1411
+ "shard": 5,
1412
+ "offset": 13258240,
1413
+ "size": 368640,
1414
+ "shape": [
1415
+ 640,
1416
+ 1024
1417
+ ],
1418
+ "dtype": "Q4_K_M",
1419
+ "role": "matmul",
1420
+ "layout": "row"
1421
+ },
1422
+ "model.layers.11.self_attn.q_norm.weight": {
1423
+ "shard": 5,
1424
+ "offset": 13626880,
1425
+ "size": 512,
1426
+ "shape": [
1427
+ 256
1428
+ ],
1429
+ "dtype": "BF16",
1430
+ "role": "norm"
1431
+ },
1432
+ "model.layers.11.self_attn.q_proj.weight": {
1433
+ "shard": 5,
1434
+ "offset": 13627392,
1435
+ "size": 442368,
1436
+ "shape": [
1437
+ 1024,
1438
+ 640
1439
+ ],
1440
+ "dtype": "Q4_K_M",
1441
+ "role": "matmul",
1442
+ "layout": "row"
1443
+ },
1444
+ "model.layers.11.self_attn.v_proj.weight": {
1445
+ "shard": 5,
1446
+ "offset": 14069760,
1447
+ "size": 110592,
1448
+ "shape": [
1449
+ 256,
1450
+ 640
1451
+ ],
1452
+ "dtype": "Q4_K_M",
1453
+ "role": "matmul",
1454
+ "layout": "row"
1455
+ },
1456
+ "model.layers.12.input_layernorm.weight": {
1457
+ "shard": 5,
1458
+ "offset": 14180352,
1459
+ "size": 1280,
1460
+ "shape": [
1461
+ 640
1462
+ ],
1463
+ "dtype": "BF16",
1464
+ "role": "norm"
1465
+ },
1466
+ "model.layers.12.mlp.down_proj.weight": {
1467
+ "shard": 5,
1468
+ "offset": 14181632,
1469
+ "size": 737280,
1470
+ "shape": [
1471
+ 640,
1472
+ 2048
1473
+ ],
1474
+ "dtype": "Q4_K_M",
1475
+ "role": "matmul",
1476
+ "layout": "row"
1477
+ },
1478
+ "model.layers.12.mlp.gate_proj.weight": {
1479
+ "shard": 5,
1480
+ "offset": 14918912,
1481
+ "size": 884736,
1482
+ "shape": [
1483
+ 2048,
1484
+ 640
1485
+ ],
1486
+ "dtype": "Q4_K_M",
1487
+ "role": "matmul",
1488
+ "layout": "row"
1489
+ },
1490
+ "model.layers.12.mlp.up_proj.weight": {
1491
+ "shard": 5,
1492
+ "offset": 15803648,
1493
+ "size": 884736,
1494
+ "shape": [
1495
+ 2048,
1496
+ 640
1497
+ ],
1498
+ "dtype": "Q4_K_M",
1499
+ "role": "matmul",
1500
+ "layout": "row"
1501
+ },
1502
+ "model.layers.12.post_attention_layernorm.weight": {
1503
+ "shard": 5,
1504
+ "offset": 16688384,
1505
+ "size": 1280,
1506
+ "shape": [
1507
+ 640
1508
+ ],
1509
+ "dtype": "BF16",
1510
+ "role": "norm"
1511
+ },
1512
+ "model.layers.12.post_feedforward_layernorm.weight": {
1513
+ "shard": 5,
1514
+ "offset": 16689664,
1515
+ "size": 1280,
1516
+ "shape": [
1517
+ 640
1518
+ ],
1519
+ "dtype": "BF16",
1520
+ "role": "norm"
1521
+ },
1522
+ "model.layers.12.pre_feedforward_layernorm.weight": {
1523
+ "shard": 5,
1524
+ "offset": 16690944,
1525
+ "size": 1280,
1526
+ "shape": [
1527
+ 640
1528
+ ],
1529
+ "dtype": "BF16",
1530
+ "role": "norm"
1531
+ },
1532
+ "model.layers.12.self_attn.k_norm.weight": {
1533
+ "shard": 5,
1534
+ "offset": 16692224,
1535
+ "size": 512,
1536
+ "shape": [
1537
+ 256
1538
+ ],
1539
+ "dtype": "BF16",
1540
+ "role": "norm"
1541
+ },
1542
+ "model.layers.12.self_attn.k_proj.weight": {
1543
+ "shard": 5,
1544
+ "offset": 16692736,
1545
+ "size": 110592,
1546
+ "shape": [
1547
+ 256,
1548
+ 640
1549
+ ],
1550
+ "dtype": "Q4_K_M",
1551
+ "role": "matmul",
1552
+ "layout": "row"
1553
+ },
1554
+ "model.layers.12.self_attn.o_proj.weight": {
1555
+ "shard": 5,
1556
+ "offset": 16803328,
1557
+ "size": 368640,
1558
+ "shape": [
1559
+ 640,
1560
+ 1024
1561
+ ],
1562
+ "dtype": "Q4_K_M",
1563
+ "role": "matmul",
1564
+ "layout": "row"
1565
+ },
1566
+ "model.layers.12.self_attn.q_norm.weight": {
1567
+ "shard": 5,
1568
+ "offset": 17171968,
1569
+ "size": 512,
1570
+ "shape": [
1571
+ 256
1572
+ ],
1573
+ "dtype": "BF16",
1574
+ "role": "norm"
1575
+ },
1576
+ "model.layers.12.self_attn.q_proj.weight": {
1577
+ "shard": 5,
1578
+ "offset": 17172480,
1579
+ "size": 442368,
1580
+ "shape": [
1581
+ 1024,
1582
+ 640
1583
+ ],
1584
+ "dtype": "Q4_K_M",
1585
+ "role": "matmul",
1586
+ "layout": "row"
1587
+ },
1588
+ "model.layers.12.self_attn.v_proj.weight": {
1589
+ "shard": 5,
1590
+ "offset": 17614848,
1591
+ "size": 110592,
1592
+ "shape": [
1593
+ 256,
1594
+ 640
1595
+ ],
1596
+ "dtype": "Q4_K_M",
1597
+ "role": "matmul",
1598
+ "layout": "row"
1599
+ },
1600
+ "model.layers.13.input_layernorm.weight": {
1601
+ "shard": 5,
1602
+ "offset": 17725440,
1603
+ "size": 1280,
1604
+ "shape": [
1605
+ 640
1606
+ ],
1607
+ "dtype": "BF16",
1608
+ "role": "norm"
1609
+ },
1610
+ "model.layers.13.mlp.down_proj.weight": {
1611
+ "shard": 5,
1612
+ "offset": 17726720,
1613
+ "size": 737280,
1614
+ "shape": [
1615
+ 640,
1616
+ 2048
1617
+ ],
1618
+ "dtype": "Q4_K_M",
1619
+ "role": "matmul",
1620
+ "layout": "row"
1621
+ },
1622
+ "model.layers.13.mlp.gate_proj.weight": {
1623
+ "shard": 5,
1624
+ "offset": 18464000,
1625
+ "size": 884736,
1626
+ "shape": [
1627
+ 2048,
1628
+ 640
1629
+ ],
1630
+ "dtype": "Q4_K_M",
1631
+ "role": "matmul",
1632
+ "layout": "row"
1633
+ },
1634
+ "model.layers.13.mlp.up_proj.weight": {
1635
+ "shard": 5,
1636
+ "offset": 19348736,
1637
+ "size": 884736,
1638
+ "shape": [
1639
+ 2048,
1640
+ 640
1641
+ ],
1642
+ "dtype": "Q4_K_M",
1643
+ "role": "matmul",
1644
+ "layout": "row"
1645
+ },
1646
+ "model.layers.13.post_attention_layernorm.weight": {
1647
+ "shard": 5,
1648
+ "offset": 20233472,
1649
+ "size": 1280,
1650
+ "shape": [
1651
+ 640
1652
+ ],
1653
+ "dtype": "BF16",
1654
+ "role": "norm"
1655
+ },
1656
+ "model.layers.13.post_feedforward_layernorm.weight": {
1657
+ "shard": 5,
1658
+ "offset": 20234752,
1659
+ "size": 1280,
1660
+ "shape": [
1661
+ 640
1662
+ ],
1663
+ "dtype": "BF16",
1664
+ "role": "norm"
1665
+ },
1666
+ "model.layers.13.pre_feedforward_layernorm.weight": {
1667
+ "shard": 5,
1668
+ "offset": 20236032,
1669
+ "size": 1280,
1670
+ "shape": [
1671
+ 640
1672
+ ],
1673
+ "dtype": "BF16",
1674
+ "role": "norm"
1675
+ },
1676
+ "model.layers.13.self_attn.k_norm.weight": {
1677
+ "shard": 5,
1678
+ "offset": 20237312,
1679
+ "size": 512,
1680
+ "shape": [
1681
+ 256
1682
+ ],
1683
+ "dtype": "BF16",
1684
+ "role": "norm"
1685
+ },
1686
+ "model.layers.13.self_attn.k_proj.weight": {
1687
+ "shard": 5,
1688
+ "offset": 20237824,
1689
+ "size": 110592,
1690
+ "shape": [
1691
+ 256,
1692
+ 640
1693
+ ],
1694
+ "dtype": "Q4_K_M",
1695
+ "role": "matmul",
1696
+ "layout": "row"
1697
+ },
1698
+ "model.layers.13.self_attn.o_proj.weight": {
1699
+ "shard": 5,
1700
+ "offset": 20348416,
1701
+ "size": 368640,
1702
+ "shape": [
1703
+ 640,
1704
+ 1024
1705
+ ],
1706
+ "dtype": "Q4_K_M",
1707
+ "role": "matmul",
1708
+ "layout": "row"
1709
+ },
1710
+ "model.layers.13.self_attn.q_norm.weight": {
1711
+ "shard": 5,
1712
+ "offset": 20717056,
1713
+ "size": 512,
1714
+ "shape": [
1715
+ 256
1716
+ ],
1717
+ "dtype": "BF16",
1718
+ "role": "norm"
1719
+ },
1720
+ "model.layers.13.self_attn.q_proj.weight": {
1721
+ "shard": 5,
1722
+ "offset": 20717568,
1723
+ "size": 442368,
1724
+ "shape": [
1725
+ 1024,
1726
+ 640
1727
+ ],
1728
+ "dtype": "Q4_K_M",
1729
+ "role": "matmul",
1730
+ "layout": "row"
1731
+ },
1732
+ "model.layers.13.self_attn.v_proj.weight": {
1733
+ "shard": 5,
1734
+ "offset": 21159936,
1735
+ "size": 110592,
1736
+ "shape": [
1737
+ 256,
1738
+ 640
1739
+ ],
1740
+ "dtype": "Q4_K_M",
1741
+ "role": "matmul",
1742
+ "layout": "row"
1743
+ },
1744
+ "model.layers.14.input_layernorm.weight": {
1745
+ "shard": 5,
1746
+ "offset": 21270528,
1747
+ "size": 1280,
1748
+ "shape": [
1749
+ 640
1750
+ ],
1751
+ "dtype": "BF16",
1752
+ "role": "norm"
1753
+ },
1754
+ "model.layers.14.mlp.down_proj.weight": {
1755
+ "shard": 5,
1756
+ "offset": 21271808,
1757
+ "size": 737280,
1758
+ "shape": [
1759
+ 640,
1760
+ 2048
1761
+ ],
1762
+ "dtype": "Q4_K_M",
1763
+ "role": "matmul",
1764
+ "layout": "row"
1765
+ },
1766
+ "model.layers.14.mlp.gate_proj.weight": {
1767
+ "shard": 5,
1768
+ "offset": 22009088,
1769
+ "size": 884736,
1770
+ "shape": [
1771
+ 2048,
1772
+ 640
1773
+ ],
1774
+ "dtype": "Q4_K_M",
1775
+ "role": "matmul",
1776
+ "layout": "row"
1777
+ },
1778
+ "model.layers.14.mlp.up_proj.weight": {
1779
+ "shard": 5,
1780
+ "offset": 22893824,
1781
+ "size": 884736,
1782
+ "shape": [
1783
+ 2048,
1784
+ 640
1785
+ ],
1786
+ "dtype": "Q4_K_M",
1787
+ "role": "matmul",
1788
+ "layout": "row"
1789
+ },
1790
+ "model.layers.14.post_attention_layernorm.weight": {
1791
+ "shard": 5,
1792
+ "offset": 23778560,
1793
+ "size": 1280,
1794
+ "shape": [
1795
+ 640
1796
+ ],
1797
+ "dtype": "BF16",
1798
+ "role": "norm"
1799
+ },
1800
+ "model.layers.14.post_feedforward_layernorm.weight": {
1801
+ "shard": 5,
1802
+ "offset": 23779840,
1803
+ "size": 1280,
1804
+ "shape": [
1805
+ 640
1806
+ ],
1807
+ "dtype": "BF16",
1808
+ "role": "norm"
1809
+ },
1810
+ "model.layers.14.pre_feedforward_layernorm.weight": {
1811
+ "shard": 5,
1812
+ "offset": 23781120,
1813
+ "size": 1280,
1814
+ "shape": [
1815
+ 640
1816
+ ],
1817
+ "dtype": "BF16",
1818
+ "role": "norm"
1819
+ },
1820
+ "model.layers.14.self_attn.k_norm.weight": {
1821
+ "shard": 5,
1822
+ "offset": 23782400,
1823
+ "size": 512,
1824
+ "shape": [
1825
+ 256
1826
+ ],
1827
+ "dtype": "BF16",
1828
+ "role": "norm"
1829
+ },
1830
+ "model.layers.14.self_attn.k_proj.weight": {
1831
+ "shard": 5,
1832
+ "offset": 23782912,
1833
+ "size": 110592,
1834
+ "shape": [
1835
+ 256,
1836
+ 640
1837
+ ],
1838
+ "dtype": "Q4_K_M",
1839
+ "role": "matmul",
1840
+ "layout": "row"
1841
+ },
1842
+ "model.layers.14.self_attn.o_proj.weight": {
1843
+ "shard": 5,
1844
+ "offset": 23893504,
1845
+ "size": 368640,
1846
+ "shape": [
1847
+ 640,
1848
+ 1024
1849
+ ],
1850
+ "dtype": "Q4_K_M",
1851
+ "role": "matmul",
1852
+ "layout": "row"
1853
+ },
1854
+ "model.layers.14.self_attn.q_norm.weight": {
1855
+ "shard": 5,
1856
+ "offset": 24262144,
1857
+ "size": 512,
1858
+ "shape": [
1859
+ 256
1860
+ ],
1861
+ "dtype": "BF16",
1862
+ "role": "norm"
1863
+ },
1864
+ "model.layers.14.self_attn.q_proj.weight": {
1865
+ "shard": 5,
1866
+ "offset": 24262656,
1867
+ "size": 442368,
1868
+ "shape": [
1869
+ 1024,
1870
+ 640
1871
+ ],
1872
+ "dtype": "Q4_K_M",
1873
+ "role": "matmul",
1874
+ "layout": "row"
1875
+ },
1876
+ "model.layers.14.self_attn.v_proj.weight": {
1877
+ "shard": 5,
1878
+ "offset": 24705024,
1879
+ "size": 110592,
1880
+ "shape": [
1881
+ 256,
1882
+ 640
1883
+ ],
1884
+ "dtype": "Q4_K_M",
1885
+ "role": "matmul",
1886
+ "layout": "row"
1887
+ },
1888
+ "model.layers.15.input_layernorm.weight": {
1889
+ "shard": 5,
1890
+ "offset": 24815616,
1891
+ "size": 1280,
1892
+ "shape": [
1893
+ 640
1894
+ ],
1895
+ "dtype": "BF16",
1896
+ "role": "norm"
1897
+ },
1898
+ "model.layers.15.mlp.down_proj.weight": {
1899
+ "shard": 5,
1900
+ "offset": 24816896,
1901
+ "size": 737280,
1902
+ "shape": [
1903
+ 640,
1904
+ 2048
1905
+ ],
1906
+ "dtype": "Q4_K_M",
1907
+ "role": "matmul",
1908
+ "layout": "row"
1909
+ },
1910
+ "model.layers.15.mlp.gate_proj.weight": {
1911
+ "shard": 5,
1912
+ "offset": 25554176,
1913
+ "size": 884736,
1914
+ "shape": [
1915
+ 2048,
1916
+ 640
1917
+ ],
1918
+ "dtype": "Q4_K_M",
1919
+ "role": "matmul",
1920
+ "layout": "row"
1921
+ },
1922
+ "model.layers.15.mlp.up_proj.weight": {
1923
+ "shard": 5,
1924
+ "offset": 26438912,
1925
+ "size": 884736,
1926
+ "shape": [
1927
+ 2048,
1928
+ 640
1929
+ ],
1930
+ "dtype": "Q4_K_M",
1931
+ "role": "matmul",
1932
+ "layout": "row"
1933
+ },
1934
+ "model.layers.15.post_attention_layernorm.weight": {
1935
+ "shard": 5,
1936
+ "offset": 27323648,
1937
+ "size": 1280,
1938
+ "shape": [
1939
+ 640
1940
+ ],
1941
+ "dtype": "BF16",
1942
+ "role": "norm"
1943
+ },
1944
+ "model.layers.15.post_feedforward_layernorm.weight": {
1945
+ "shard": 5,
1946
+ "offset": 27324928,
1947
+ "size": 1280,
1948
+ "shape": [
1949
+ 640
1950
+ ],
1951
+ "dtype": "BF16",
1952
+ "role": "norm"
1953
+ },
1954
+ "model.layers.15.pre_feedforward_layernorm.weight": {
1955
+ "shard": 5,
1956
+ "offset": 27326208,
1957
+ "size": 1280,
1958
+ "shape": [
1959
+ 640
1960
+ ],
1961
+ "dtype": "BF16",
1962
+ "role": "norm"
1963
+ },
1964
+ "model.layers.15.self_attn.k_norm.weight": {
1965
+ "shard": 5,
1966
+ "offset": 27327488,
1967
+ "size": 512,
1968
+ "shape": [
1969
+ 256
1970
+ ],
1971
+ "dtype": "BF16",
1972
+ "role": "norm"
1973
+ },
1974
+ "model.layers.15.self_attn.k_proj.weight": {
1975
+ "shard": 5,
1976
+ "offset": 27328000,
1977
+ "size": 110592,
1978
+ "shape": [
1979
+ 256,
1980
+ 640
1981
+ ],
1982
+ "dtype": "Q4_K_M",
1983
+ "role": "matmul",
1984
+ "layout": "row"
1985
+ },
1986
+ "model.layers.15.self_attn.o_proj.weight": {
1987
+ "shard": 5,
1988
+ "offset": 27438592,
1989
+ "size": 368640,
1990
+ "shape": [
1991
+ 640,
1992
+ 1024
1993
+ ],
1994
+ "dtype": "Q4_K_M",
1995
+ "role": "matmul",
1996
+ "layout": "row"
1997
+ },
1998
+ "model.layers.15.self_attn.q_norm.weight": {
1999
+ "shard": 5,
2000
+ "offset": 27807232,
2001
+ "size": 512,
2002
+ "shape": [
2003
+ 256
2004
+ ],
2005
+ "dtype": "BF16",
2006
+ "role": "norm"
2007
+ },
2008
+ "model.layers.15.self_attn.q_proj.weight": {
2009
+ "shard": 5,
2010
+ "offset": 27807744,
2011
+ "size": 442368,
2012
+ "shape": [
2013
+ 1024,
2014
+ 640
2015
+ ],
2016
+ "dtype": "Q4_K_M",
2017
+ "role": "matmul",
2018
+ "layout": "row"
2019
+ },
2020
+ "model.layers.15.self_attn.v_proj.weight": {
2021
+ "shard": 5,
2022
+ "offset": 28250112,
2023
+ "size": 110592,
2024
+ "shape": [
2025
+ 256,
2026
+ 640
2027
+ ],
2028
+ "dtype": "Q4_K_M",
2029
+ "role": "matmul",
2030
+ "layout": "row"
2031
+ },
2032
+ "model.layers.16.input_layernorm.weight": {
2033
+ "shard": 5,
2034
+ "offset": 28360704,
2035
+ "size": 1280,
2036
+ "shape": [
2037
+ 640
2038
+ ],
2039
+ "dtype": "BF16",
2040
+ "role": "norm"
2041
+ },
2042
+ "model.layers.16.mlp.down_proj.weight": {
2043
+ "shard": 5,
2044
+ "offset": 28361984,
2045
+ "size": 737280,
2046
+ "shape": [
2047
+ 640,
2048
+ 2048
2049
+ ],
2050
+ "dtype": "Q4_K_M",
2051
+ "role": "matmul",
2052
+ "layout": "row"
2053
+ },
2054
+ "model.layers.16.mlp.gate_proj.weight": {
2055
+ "shard": 5,
2056
+ "offset": 29099264,
2057
+ "size": 884736,
2058
+ "shape": [
2059
+ 2048,
2060
+ 640
2061
+ ],
2062
+ "dtype": "Q4_K_M",
2063
+ "role": "matmul",
2064
+ "layout": "row"
2065
+ },
2066
+ "model.layers.16.mlp.up_proj.weight": {
2067
+ "shard": 5,
2068
+ "offset": 29984000,
2069
+ "size": 884736,
2070
+ "shape": [
2071
+ 2048,
2072
+ 640
2073
+ ],
2074
+ "dtype": "Q4_K_M",
2075
+ "role": "matmul",
2076
+ "layout": "row"
2077
+ },
2078
+ "model.layers.16.post_attention_layernorm.weight": {
2079
+ "shard": 5,
2080
+ "offset": 30868736,
2081
+ "size": 1280,
2082
+ "shape": [
2083
+ 640
2084
+ ],
2085
+ "dtype": "BF16",
2086
+ "role": "norm"
2087
+ },
2088
+ "model.layers.16.post_feedforward_layernorm.weight": {
2089
+ "shard": 5,
2090
+ "offset": 30870016,
2091
+ "size": 1280,
2092
+ "shape": [
2093
+ 640
2094
+ ],
2095
+ "dtype": "BF16",
2096
+ "role": "norm"
2097
+ },
2098
+ "model.layers.16.pre_feedforward_layernorm.weight": {
2099
+ "shard": 5,
2100
+ "offset": 30871296,
2101
+ "size": 1280,
2102
+ "shape": [
2103
+ 640
2104
+ ],
2105
+ "dtype": "BF16",
2106
+ "role": "norm"
2107
+ },
2108
+ "model.layers.16.self_attn.k_norm.weight": {
2109
+ "shard": 5,
2110
+ "offset": 30872576,
2111
+ "size": 512,
2112
+ "shape": [
2113
+ 256
2114
+ ],
2115
+ "dtype": "BF16",
2116
+ "role": "norm"
2117
+ },
2118
+ "model.layers.16.self_attn.k_proj.weight": {
2119
+ "shard": 5,
2120
+ "offset": 30873088,
2121
+ "size": 110592,
2122
+ "shape": [
2123
+ 256,
2124
+ 640
2125
+ ],
2126
+ "dtype": "Q4_K_M",
2127
+ "role": "matmul",
2128
+ "layout": "row"
2129
+ },
2130
+ "model.layers.16.self_attn.o_proj.weight": {
2131
+ "shard": 5,
2132
+ "offset": 30983680,
2133
+ "size": 368640,
2134
+ "shape": [
2135
+ 640,
2136
+ 1024
2137
+ ],
2138
+ "dtype": "Q4_K_M",
2139
+ "role": "matmul",
2140
+ "layout": "row"
2141
+ },
2142
+ "model.layers.16.self_attn.q_norm.weight": {
2143
+ "shard": 5,
2144
+ "offset": 31352320,
2145
+ "size": 512,
2146
+ "shape": [
2147
+ 256
2148
+ ],
2149
+ "dtype": "BF16",
2150
+ "role": "norm"
2151
+ },
2152
+ "model.layers.16.self_attn.q_proj.weight": {
2153
+ "shard": 5,
2154
+ "offset": 31352832,
2155
+ "size": 442368,
2156
+ "shape": [
2157
+ 1024,
2158
+ 640
2159
+ ],
2160
+ "dtype": "Q4_K_M",
2161
+ "role": "matmul",
2162
+ "layout": "row"
2163
+ },
2164
+ "model.layers.16.self_attn.v_proj.weight": {
2165
+ "shard": 5,
2166
+ "offset": 31795200,
2167
+ "size": 110592,
2168
+ "shape": [
2169
+ 256,
2170
+ 640
2171
+ ],
2172
+ "dtype": "Q4_K_M",
2173
+ "role": "matmul",
2174
+ "layout": "row"
2175
+ },
2176
+ "model.layers.17.input_layernorm.weight": {
2177
+ "shard": 5,
2178
+ "offset": 31905792,
2179
+ "size": 1280,
2180
+ "shape": [
2181
+ 640
2182
+ ],
2183
+ "dtype": "BF16",
2184
+ "role": "norm"
2185
+ },
2186
+ "model.layers.17.mlp.down_proj.weight": {
2187
+ "shard": 5,
2188
+ "offset": 31907072,
2189
+ "size": 737280,
2190
+ "shape": [
2191
+ 640,
2192
+ 2048
2193
+ ],
2194
+ "dtype": "Q4_K_M",
2195
+ "role": "matmul",
2196
+ "layout": "row"
2197
+ },
2198
+ "model.layers.17.mlp.gate_proj.weight": {
2199
+ "shard": 5,
2200
+ "offset": 32644352,
2201
+ "size": 884736,
2202
+ "shape": [
2203
+ 2048,
2204
+ 640
2205
+ ],
2206
+ "dtype": "Q4_K_M",
2207
+ "role": "matmul",
2208
+ "layout": "row"
2209
+ },
2210
+ "model.layers.17.mlp.up_proj.weight": {
2211
+ "shard": 5,
2212
+ "offset": 33529088,
2213
+ "size": 884736,
2214
+ "shape": [
2215
+ 2048,
2216
+ 640
2217
+ ],
2218
+ "dtype": "Q4_K_M",
2219
+ "role": "matmul",
2220
+ "layout": "row"
2221
+ },
2222
+ "model.layers.17.post_attention_layernorm.weight": {
2223
+ "shard": 5,
2224
+ "offset": 34413824,
2225
+ "size": 1280,
2226
+ "shape": [
2227
+ 640
2228
+ ],
2229
+ "dtype": "BF16",
2230
+ "role": "norm"
2231
+ },
2232
+ "model.layers.17.post_feedforward_layernorm.weight": {
2233
+ "shard": 5,
2234
+ "offset": 34415104,
2235
+ "size": 1280,
2236
+ "shape": [
2237
+ 640
2238
+ ],
2239
+ "dtype": "BF16",
2240
+ "role": "norm"
2241
+ },
2242
+ "model.layers.17.pre_feedforward_layernorm.weight": {
2243
+ "shard": 5,
2244
+ "offset": 34416384,
2245
+ "size": 1280,
2246
+ "shape": [
2247
+ 640
2248
+ ],
2249
+ "dtype": "BF16",
2250
+ "role": "norm"
2251
+ },
2252
+ "model.layers.17.self_attn.k_norm.weight": {
2253
+ "shard": 5,
2254
+ "offset": 34417664,
2255
+ "size": 512,
2256
+ "shape": [
2257
+ 256
2258
+ ],
2259
+ "dtype": "BF16",
2260
+ "role": "norm"
2261
+ },
2262
+ "model.layers.17.self_attn.k_proj.weight": {
2263
+ "shard": 5,
2264
+ "offset": 34418176,
2265
+ "size": 110592,
2266
+ "shape": [
2267
+ 256,
2268
+ 640
2269
+ ],
2270
+ "dtype": "Q4_K_M",
2271
+ "role": "matmul",
2272
+ "layout": "row"
2273
+ },
2274
+ "model.layers.17.self_attn.o_proj.weight": {
2275
+ "shard": 5,
2276
+ "offset": 34528768,
2277
+ "size": 368640,
2278
+ "shape": [
2279
+ 640,
2280
+ 1024
2281
+ ],
2282
+ "dtype": "Q4_K_M",
2283
+ "role": "matmul",
2284
+ "layout": "row"
2285
+ },
2286
+ "model.layers.17.self_attn.q_norm.weight": {
2287
+ "shard": 5,
2288
+ "offset": 34897408,
2289
+ "size": 512,
2290
+ "shape": [
2291
+ 256
2292
+ ],
2293
+ "dtype": "BF16",
2294
+ "role": "norm"
2295
+ },
2296
+ "model.layers.17.self_attn.q_proj.weight": {
2297
+ "shard": 5,
2298
+ "offset": 34897920,
2299
+ "size": 442368,
2300
+ "shape": [
2301
+ 1024,
2302
+ 640
2303
+ ],
2304
+ "dtype": "Q4_K_M",
2305
+ "role": "matmul",
2306
+ "layout": "row"
2307
+ },
2308
+ "model.layers.17.self_attn.v_proj.weight": {
2309
+ "shard": 5,
2310
+ "offset": 35340288,
2311
+ "size": 110592,
2312
+ "shape": [
2313
+ 256,
2314
+ 640
2315
+ ],
2316
+ "dtype": "Q4_K_M",
2317
+ "role": "matmul",
2318
+ "layout": "row"
2319
+ },
2320
+ "model.layers.2.input_layernorm.weight": {
2321
+ "shard": 5,
2322
+ "offset": 35450880,
2323
+ "size": 1280,
2324
+ "shape": [
2325
+ 640
2326
+ ],
2327
+ "dtype": "BF16",
2328
+ "role": "norm"
2329
+ },
2330
+ "model.layers.2.mlp.down_proj.weight": {
2331
+ "shard": 5,
2332
+ "offset": 35452160,
2333
+ "size": 737280,
2334
+ "shape": [
2335
+ 640,
2336
+ 2048
2337
+ ],
2338
+ "dtype": "Q4_K_M",
2339
+ "role": "matmul",
2340
+ "layout": "row"
2341
+ },
2342
+ "model.layers.2.mlp.gate_proj.weight": {
2343
+ "shard": 5,
2344
+ "offset": 36189440,
2345
+ "size": 884736,
2346
+ "shape": [
2347
+ 2048,
2348
+ 640
2349
+ ],
2350
+ "dtype": "Q4_K_M",
2351
+ "role": "matmul",
2352
+ "layout": "row"
2353
+ },
2354
+ "model.layers.2.mlp.up_proj.weight": {
2355
+ "shard": 5,
2356
+ "offset": 37074176,
2357
+ "size": 884736,
2358
+ "shape": [
2359
+ 2048,
2360
+ 640
2361
+ ],
2362
+ "dtype": "Q4_K_M",
2363
+ "role": "matmul",
2364
+ "layout": "row"
2365
+ },
2366
+ "model.layers.2.post_attention_layernorm.weight": {
2367
+ "shard": 5,
2368
+ "offset": 37958912,
2369
+ "size": 1280,
2370
+ "shape": [
2371
+ 640
2372
+ ],
2373
+ "dtype": "BF16",
2374
+ "role": "norm"
2375
+ },
2376
+ "model.layers.2.post_feedforward_layernorm.weight": {
2377
+ "shard": 5,
2378
+ "offset": 37960192,
2379
+ "size": 1280,
2380
+ "shape": [
2381
+ 640
2382
+ ],
2383
+ "dtype": "BF16",
2384
+ "role": "norm"
2385
+ },
2386
+ "model.layers.2.pre_feedforward_layernorm.weight": {
2387
+ "shard": 5,
2388
+ "offset": 37961472,
2389
+ "size": 1280,
2390
+ "shape": [
2391
+ 640
2392
+ ],
2393
+ "dtype": "BF16",
2394
+ "role": "norm"
2395
+ },
2396
+ "model.layers.2.self_attn.k_norm.weight": {
2397
+ "shard": 5,
2398
+ "offset": 37962752,
2399
+ "size": 512,
2400
+ "shape": [
2401
+ 256
2402
+ ],
2403
+ "dtype": "BF16",
2404
+ "role": "norm"
2405
+ },
2406
+ "model.layers.2.self_attn.k_proj.weight": {
2407
+ "shard": 5,
2408
+ "offset": 37963264,
2409
+ "size": 110592,
2410
+ "shape": [
2411
+ 256,
2412
+ 640
2413
+ ],
2414
+ "dtype": "Q4_K_M",
2415
+ "role": "matmul",
2416
+ "layout": "row"
2417
+ },
2418
+ "model.layers.2.self_attn.o_proj.weight": {
2419
+ "shard": 5,
2420
+ "offset": 38073856,
2421
+ "size": 368640,
2422
+ "shape": [
2423
+ 640,
2424
+ 1024
2425
+ ],
2426
+ "dtype": "Q4_K_M",
2427
+ "role": "matmul",
2428
+ "layout": "row"
2429
+ },
2430
+ "model.layers.2.self_attn.q_norm.weight": {
2431
+ "shard": 5,
2432
+ "offset": 38442496,
2433
+ "size": 512,
2434
+ "shape": [
2435
+ 256
2436
+ ],
2437
+ "dtype": "BF16",
2438
+ "role": "norm"
2439
+ },
2440
+ "model.layers.2.self_attn.q_proj.weight": {
2441
+ "shard": 5,
2442
+ "offset": 38443008,
2443
+ "size": 442368,
2444
+ "shape": [
2445
+ 1024,
2446
+ 640
2447
+ ],
2448
+ "dtype": "Q4_K_M",
2449
+ "role": "matmul",
2450
+ "layout": "row"
2451
+ },
2452
+ "model.layers.2.self_attn.v_proj.weight": {
2453
+ "shard": 5,
2454
+ "offset": 38885376,
2455
+ "size": 110592,
2456
+ "shape": [
2457
+ 256,
2458
+ 640
2459
+ ],
2460
+ "dtype": "Q4_K_M",
2461
+ "role": "matmul",
2462
+ "layout": "row"
2463
+ },
2464
+ "model.layers.3.input_layernorm.weight": {
2465
+ "shard": 5,
2466
+ "offset": 38995968,
2467
+ "size": 1280,
2468
+ "shape": [
2469
+ 640
2470
+ ],
2471
+ "dtype": "BF16",
2472
+ "role": "norm"
2473
+ },
2474
+ "model.layers.3.mlp.down_proj.weight": {
2475
+ "shard": 5,
2476
+ "offset": 38997248,
2477
+ "size": 737280,
2478
+ "shape": [
2479
+ 640,
2480
+ 2048
2481
+ ],
2482
+ "dtype": "Q4_K_M",
2483
+ "role": "matmul",
2484
+ "layout": "row"
2485
+ },
2486
+ "model.layers.3.mlp.gate_proj.weight": {
2487
+ "shard": 5,
2488
+ "offset": 39734528,
2489
+ "size": 884736,
2490
+ "shape": [
2491
+ 2048,
2492
+ 640
2493
+ ],
2494
+ "dtype": "Q4_K_M",
2495
+ "role": "matmul",
2496
+ "layout": "row"
2497
+ },
2498
+ "model.layers.3.mlp.up_proj.weight": {
2499
+ "shard": 5,
2500
+ "offset": 40619264,
2501
+ "size": 884736,
2502
+ "shape": [
2503
+ 2048,
2504
+ 640
2505
+ ],
2506
+ "dtype": "Q4_K_M",
2507
+ "role": "matmul",
2508
+ "layout": "row"
2509
+ },
2510
+ "model.layers.3.post_attention_layernorm.weight": {
2511
+ "shard": 5,
2512
+ "offset": 41504000,
2513
+ "size": 1280,
2514
+ "shape": [
2515
+ 640
2516
+ ],
2517
+ "dtype": "BF16",
2518
+ "role": "norm"
2519
+ },
2520
+ "model.layers.3.post_feedforward_layernorm.weight": {
2521
+ "shard": 5,
2522
+ "offset": 41505280,
2523
+ "size": 1280,
2524
+ "shape": [
2525
+ 640
2526
+ ],
2527
+ "dtype": "BF16",
2528
+ "role": "norm"
2529
+ },
2530
+ "model.layers.3.pre_feedforward_layernorm.weight": {
2531
+ "shard": 5,
2532
+ "offset": 41506560,
2533
+ "size": 1280,
2534
+ "shape": [
2535
+ 640
2536
+ ],
2537
+ "dtype": "BF16",
2538
+ "role": "norm"
2539
+ },
2540
+ "model.layers.3.self_attn.k_norm.weight": {
2541
+ "shard": 5,
2542
+ "offset": 41507840,
2543
+ "size": 512,
2544
+ "shape": [
2545
+ 256
2546
+ ],
2547
+ "dtype": "BF16",
2548
+ "role": "norm"
2549
+ },
2550
+ "model.layers.3.self_attn.k_proj.weight": {
2551
+ "shard": 5,
2552
+ "offset": 41508352,
2553
+ "size": 110592,
2554
+ "shape": [
2555
+ 256,
2556
+ 640
2557
+ ],
2558
+ "dtype": "Q4_K_M",
2559
+ "role": "matmul",
2560
+ "layout": "row"
2561
+ },
2562
+ "model.layers.3.self_attn.o_proj.weight": {
2563
+ "shard": 5,
2564
+ "offset": 41618944,
2565
+ "size": 368640,
2566
+ "shape": [
2567
+ 640,
2568
+ 1024
2569
+ ],
2570
+ "dtype": "Q4_K_M",
2571
+ "role": "matmul",
2572
+ "layout": "row"
2573
+ },
2574
+ "model.layers.3.self_attn.q_norm.weight": {
2575
+ "shard": 5,
2576
+ "offset": 41987584,
2577
+ "size": 512,
2578
+ "shape": [
2579
+ 256
2580
+ ],
2581
+ "dtype": "BF16",
2582
+ "role": "norm"
2583
+ },
2584
+ "model.layers.3.self_attn.q_proj.weight": {
2585
+ "shard": 5,
2586
+ "offset": 41988096,
2587
+ "size": 442368,
2588
+ "shape": [
2589
+ 1024,
2590
+ 640
2591
+ ],
2592
+ "dtype": "Q4_K_M",
2593
+ "role": "matmul",
2594
+ "layout": "row"
2595
+ },
2596
+ "model.layers.3.self_attn.v_proj.weight": {
2597
+ "shard": 5,
2598
+ "offset": 42430464,
2599
+ "size": 110592,
2600
+ "shape": [
2601
+ 256,
2602
+ 640
2603
+ ],
2604
+ "dtype": "Q4_K_M",
2605
+ "role": "matmul",
2606
+ "layout": "row"
2607
+ },
2608
+ "model.layers.4.input_layernorm.weight": {
2609
+ "shard": 5,
2610
+ "offset": 42541056,
2611
+ "size": 1280,
2612
+ "shape": [
2613
+ 640
2614
+ ],
2615
+ "dtype": "BF16",
2616
+ "role": "norm"
2617
+ },
2618
+ "model.layers.4.mlp.down_proj.weight": {
2619
+ "shard": 5,
2620
+ "offset": 42542336,
2621
+ "size": 737280,
2622
+ "shape": [
2623
+ 640,
2624
+ 2048
2625
+ ],
2626
+ "dtype": "Q4_K_M",
2627
+ "role": "matmul",
2628
+ "layout": "row"
2629
+ },
2630
+ "model.layers.4.mlp.gate_proj.weight": {
2631
+ "shard": 5,
2632
+ "offset": 43279616,
2633
+ "size": 884736,
2634
+ "shape": [
2635
+ 2048,
2636
+ 640
2637
+ ],
2638
+ "dtype": "Q4_K_M",
2639
+ "role": "matmul",
2640
+ "layout": "row"
2641
+ },
2642
+ "model.layers.4.mlp.up_proj.weight": {
2643
+ "shard": 5,
2644
+ "offset": 44164352,
2645
+ "size": 884736,
2646
+ "shape": [
2647
+ 2048,
2648
+ 640
2649
+ ],
2650
+ "dtype": "Q4_K_M",
2651
+ "role": "matmul",
2652
+ "layout": "row"
2653
+ },
2654
+ "model.layers.4.post_attention_layernorm.weight": {
2655
+ "shard": 5,
2656
+ "offset": 45049088,
2657
+ "size": 1280,
2658
+ "shape": [
2659
+ 640
2660
+ ],
2661
+ "dtype": "BF16",
2662
+ "role": "norm"
2663
+ },
2664
+ "model.layers.4.post_feedforward_layernorm.weight": {
2665
+ "shard": 5,
2666
+ "offset": 45050368,
2667
+ "size": 1280,
2668
+ "shape": [
2669
+ 640
2670
+ ],
2671
+ "dtype": "BF16",
2672
+ "role": "norm"
2673
+ },
2674
+ "model.layers.4.pre_feedforward_layernorm.weight": {
2675
+ "shard": 5,
2676
+ "offset": 45051648,
2677
+ "size": 1280,
2678
+ "shape": [
2679
+ 640
2680
+ ],
2681
+ "dtype": "BF16",
2682
+ "role": "norm"
2683
+ },
2684
+ "model.layers.4.self_attn.k_norm.weight": {
2685
+ "shard": 5,
2686
+ "offset": 45052928,
2687
+ "size": 512,
2688
+ "shape": [
2689
+ 256
2690
+ ],
2691
+ "dtype": "BF16",
2692
+ "role": "norm"
2693
+ },
2694
+ "model.layers.4.self_attn.k_proj.weight": {
2695
+ "shard": 5,
2696
+ "offset": 45053440,
2697
+ "size": 110592,
2698
+ "shape": [
2699
+ 256,
2700
+ 640
2701
+ ],
2702
+ "dtype": "Q4_K_M",
2703
+ "role": "matmul",
2704
+ "layout": "row"
2705
+ },
2706
+ "model.layers.4.self_attn.o_proj.weight": {
2707
+ "shard": 5,
2708
+ "offset": 45164032,
2709
+ "size": 368640,
2710
+ "shape": [
2711
+ 640,
2712
+ 1024
2713
+ ],
2714
+ "dtype": "Q4_K_M",
2715
+ "role": "matmul",
2716
+ "layout": "row"
2717
+ },
2718
+ "model.layers.4.self_attn.q_norm.weight": {
2719
+ "shard": 5,
2720
+ "offset": 45532672,
2721
+ "size": 512,
2722
+ "shape": [
2723
+ 256
2724
+ ],
2725
+ "dtype": "BF16",
2726
+ "role": "norm"
2727
+ },
2728
+ "model.layers.4.self_attn.q_proj.weight": {
2729
+ "shard": 5,
2730
+ "offset": 45533184,
2731
+ "size": 442368,
2732
+ "shape": [
2733
+ 1024,
2734
+ 640
2735
+ ],
2736
+ "dtype": "Q4_K_M",
2737
+ "role": "matmul",
2738
+ "layout": "row"
2739
+ },
2740
+ "model.layers.4.self_attn.v_proj.weight": {
2741
+ "shard": 5,
2742
+ "offset": 45975552,
2743
+ "size": 110592,
2744
+ "shape": [
2745
+ 256,
2746
+ 640
2747
+ ],
2748
+ "dtype": "Q4_K_M",
2749
+ "role": "matmul",
2750
+ "layout": "row"
2751
+ },
2752
+ "model.layers.5.input_layernorm.weight": {
2753
+ "shard": 5,
2754
+ "offset": 46086144,
2755
+ "size": 1280,
2756
+ "shape": [
2757
+ 640
2758
+ ],
2759
+ "dtype": "BF16",
2760
+ "role": "norm"
2761
+ },
2762
+ "model.layers.5.mlp.down_proj.weight": {
2763
+ "shard": 5,
2764
+ "offset": 46087424,
2765
+ "size": 737280,
2766
+ "shape": [
2767
+ 640,
2768
+ 2048
2769
+ ],
2770
+ "dtype": "Q4_K_M",
2771
+ "role": "matmul",
2772
+ "layout": "row"
2773
+ },
2774
+ "model.layers.5.mlp.gate_proj.weight": {
2775
+ "shard": 5,
2776
+ "offset": 46824704,
2777
+ "size": 884736,
2778
+ "shape": [
2779
+ 2048,
2780
+ 640
2781
+ ],
2782
+ "dtype": "Q4_K_M",
2783
+ "role": "matmul",
2784
+ "layout": "row"
2785
+ },
2786
+ "model.layers.5.mlp.up_proj.weight": {
2787
+ "shard": 5,
2788
+ "offset": 47709440,
2789
+ "size": 884736,
2790
+ "shape": [
2791
+ 2048,
2792
+ 640
2793
+ ],
2794
+ "dtype": "Q4_K_M",
2795
+ "role": "matmul",
2796
+ "layout": "row"
2797
+ },
2798
+ "model.layers.5.post_attention_layernorm.weight": {
2799
+ "shard": 5,
2800
+ "offset": 48594176,
2801
+ "size": 1280,
2802
+ "shape": [
2803
+ 640
2804
+ ],
2805
+ "dtype": "BF16",
2806
+ "role": "norm"
2807
+ },
2808
+ "model.layers.5.post_feedforward_layernorm.weight": {
2809
+ "shard": 5,
2810
+ "offset": 48595456,
2811
+ "size": 1280,
2812
+ "shape": [
2813
+ 640
2814
+ ],
2815
+ "dtype": "BF16",
2816
+ "role": "norm"
2817
+ },
2818
+ "model.layers.5.pre_feedforward_layernorm.weight": {
2819
+ "shard": 5,
2820
+ "offset": 48596736,
2821
+ "size": 1280,
2822
+ "shape": [
2823
+ 640
2824
+ ],
2825
+ "dtype": "BF16",
2826
+ "role": "norm"
2827
+ },
2828
+ "model.layers.5.self_attn.k_norm.weight": {
2829
+ "shard": 5,
2830
+ "offset": 48598016,
2831
+ "size": 512,
2832
+ "shape": [
2833
+ 256
2834
+ ],
2835
+ "dtype": "BF16",
2836
+ "role": "norm"
2837
+ },
2838
+ "model.layers.5.self_attn.k_proj.weight": {
2839
+ "shard": 5,
2840
+ "offset": 48598528,
2841
+ "size": 110592,
2842
+ "shape": [
2843
+ 256,
2844
+ 640
2845
+ ],
2846
+ "dtype": "Q4_K_M",
2847
+ "role": "matmul",
2848
+ "layout": "row"
2849
+ },
2850
+ "model.layers.5.self_attn.o_proj.weight": {
2851
+ "shard": 5,
2852
+ "offset": 48709120,
2853
+ "size": 368640,
2854
+ "shape": [
2855
+ 640,
2856
+ 1024
2857
+ ],
2858
+ "dtype": "Q4_K_M",
2859
+ "role": "matmul",
2860
+ "layout": "row"
2861
+ },
2862
+ "model.layers.5.self_attn.q_norm.weight": {
2863
+ "shard": 5,
2864
+ "offset": 49077760,
2865
+ "size": 512,
2866
+ "shape": [
2867
+ 256
2868
+ ],
2869
+ "dtype": "BF16",
2870
+ "role": "norm"
2871
+ },
2872
+ "model.layers.5.self_attn.q_proj.weight": {
2873
+ "shard": 5,
2874
+ "offset": 49078272,
2875
+ "size": 442368,
2876
+ "shape": [
2877
+ 1024,
2878
+ 640
2879
+ ],
2880
+ "dtype": "Q4_K_M",
2881
+ "role": "matmul",
2882
+ "layout": "row"
2883
+ },
2884
+ "model.layers.5.self_attn.v_proj.weight": {
2885
+ "shard": 5,
2886
+ "offset": 49520640,
2887
+ "size": 110592,
2888
+ "shape": [
2889
+ 256,
2890
+ 640
2891
+ ],
2892
+ "dtype": "Q4_K_M",
2893
+ "role": "matmul",
2894
+ "layout": "row"
2895
+ },
2896
+ "model.layers.6.input_layernorm.weight": {
2897
+ "shard": 5,
2898
+ "offset": 49631232,
2899
+ "size": 1280,
2900
+ "shape": [
2901
+ 640
2902
+ ],
2903
+ "dtype": "BF16",
2904
+ "role": "norm"
2905
+ },
2906
+ "model.layers.6.mlp.down_proj.weight": {
2907
+ "shard": 5,
2908
+ "offset": 49632512,
2909
+ "size": 737280,
2910
+ "shape": [
2911
+ 640,
2912
+ 2048
2913
+ ],
2914
+ "dtype": "Q4_K_M",
2915
+ "role": "matmul",
2916
+ "layout": "row"
2917
+ },
2918
+ "model.layers.6.mlp.gate_proj.weight": {
2919
+ "shard": 5,
2920
+ "offset": 50369792,
2921
+ "size": 884736,
2922
+ "shape": [
2923
+ 2048,
2924
+ 640
2925
+ ],
2926
+ "dtype": "Q4_K_M",
2927
+ "role": "matmul",
2928
+ "layout": "row"
2929
+ },
2930
+ "model.layers.6.mlp.up_proj.weight": {
2931
+ "shard": 5,
2932
+ "offset": 51254528,
2933
+ "size": 884736,
2934
+ "shape": [
2935
+ 2048,
2936
+ 640
2937
+ ],
2938
+ "dtype": "Q4_K_M",
2939
+ "role": "matmul",
2940
+ "layout": "row"
2941
+ },
2942
+ "model.layers.6.post_attention_layernorm.weight": {
2943
+ "shard": 5,
2944
+ "offset": 52139264,
2945
+ "size": 1280,
2946
+ "shape": [
2947
+ 640
2948
+ ],
2949
+ "dtype": "BF16",
2950
+ "role": "norm"
2951
+ },
2952
+ "model.layers.6.post_feedforward_layernorm.weight": {
2953
+ "shard": 5,
2954
+ "offset": 52140544,
2955
+ "size": 1280,
2956
+ "shape": [
2957
+ 640
2958
+ ],
2959
+ "dtype": "BF16",
2960
+ "role": "norm"
2961
+ },
2962
+ "model.layers.6.pre_feedforward_layernorm.weight": {
2963
+ "shard": 5,
2964
+ "offset": 52141824,
2965
+ "size": 1280,
2966
+ "shape": [
2967
+ 640
2968
+ ],
2969
+ "dtype": "BF16",
2970
+ "role": "norm"
2971
+ },
2972
+ "model.layers.6.self_attn.k_norm.weight": {
2973
+ "shard": 5,
2974
+ "offset": 52143104,
2975
+ "size": 512,
2976
+ "shape": [
2977
+ 256
2978
+ ],
2979
+ "dtype": "BF16",
2980
+ "role": "norm"
2981
+ },
2982
+ "model.layers.6.self_attn.k_proj.weight": {
2983
+ "shard": 5,
2984
+ "offset": 52143616,
2985
+ "size": 110592,
2986
+ "shape": [
2987
+ 256,
2988
+ 640
2989
+ ],
2990
+ "dtype": "Q4_K_M",
2991
+ "role": "matmul",
2992
+ "layout": "row"
2993
+ },
2994
+ "model.layers.6.self_attn.o_proj.weight": {
2995
+ "shard": 5,
2996
+ "offset": 52254208,
2997
+ "size": 368640,
2998
+ "shape": [
2999
+ 640,
3000
+ 1024
3001
+ ],
3002
+ "dtype": "Q4_K_M",
3003
+ "role": "matmul",
3004
+ "layout": "row"
3005
+ },
3006
+ "model.layers.6.self_attn.q_norm.weight": {
3007
+ "shard": 5,
3008
+ "offset": 52622848,
3009
+ "size": 512,
3010
+ "shape": [
3011
+ 256
3012
+ ],
3013
+ "dtype": "BF16",
3014
+ "role": "norm"
3015
+ },
3016
+ "model.layers.6.self_attn.q_proj.weight": {
3017
+ "shard": 5,
3018
+ "offset": 52623360,
3019
+ "size": 442368,
3020
+ "shape": [
3021
+ 1024,
3022
+ 640
3023
+ ],
3024
+ "dtype": "Q4_K_M",
3025
+ "role": "matmul",
3026
+ "layout": "row"
3027
+ },
3028
+ "model.layers.6.self_attn.v_proj.weight": {
3029
+ "shard": 5,
3030
+ "offset": 53065728,
3031
+ "size": 110592,
3032
+ "shape": [
3033
+ 256,
3034
+ 640
3035
+ ],
3036
+ "dtype": "Q4_K_M",
3037
+ "role": "matmul",
3038
+ "layout": "row"
3039
+ },
3040
+ "model.layers.7.input_layernorm.weight": {
3041
+ "shard": 5,
3042
+ "offset": 53176320,
3043
+ "size": 1280,
3044
+ "shape": [
3045
+ 640
3046
+ ],
3047
+ "dtype": "BF16",
3048
+ "role": "norm"
3049
+ },
3050
+ "model.layers.7.mlp.down_proj.weight": {
3051
+ "shard": 5,
3052
+ "offset": 53177600,
3053
+ "size": 737280,
3054
+ "shape": [
3055
+ 640,
3056
+ 2048
3057
+ ],
3058
+ "dtype": "Q4_K_M",
3059
+ "role": "matmul",
3060
+ "layout": "row"
3061
+ },
3062
+ "model.layers.7.mlp.gate_proj.weight": {
3063
+ "shard": 5,
3064
+ "offset": 53914880,
3065
+ "size": 884736,
3066
+ "shape": [
3067
+ 2048,
3068
+ 640
3069
+ ],
3070
+ "dtype": "Q4_K_M",
3071
+ "role": "matmul",
3072
+ "layout": "row"
3073
+ },
3074
+ "model.layers.7.mlp.up_proj.weight": {
3075
+ "shard": 5,
3076
+ "offset": 54799616,
3077
+ "size": 884736,
3078
+ "shape": [
3079
+ 2048,
3080
+ 640
3081
+ ],
3082
+ "dtype": "Q4_K_M",
3083
+ "role": "matmul",
3084
+ "layout": "row"
3085
+ },
3086
+ "model.layers.7.post_attention_layernorm.weight": {
3087
+ "shard": 5,
3088
+ "offset": 55684352,
3089
+ "size": 1280,
3090
+ "shape": [
3091
+ 640
3092
+ ],
3093
+ "dtype": "BF16",
3094
+ "role": "norm"
3095
+ },
3096
+ "model.layers.7.post_feedforward_layernorm.weight": {
3097
+ "shard": 5,
3098
+ "offset": 55685632,
3099
+ "size": 1280,
3100
+ "shape": [
3101
+ 640
3102
+ ],
3103
+ "dtype": "BF16",
3104
+ "role": "norm"
3105
+ },
3106
+ "model.layers.7.pre_feedforward_layernorm.weight": {
3107
+ "shard": 5,
3108
+ "offset": 55686912,
3109
+ "size": 1280,
3110
+ "shape": [
3111
+ 640
3112
+ ],
3113
+ "dtype": "BF16",
3114
+ "role": "norm"
3115
+ },
3116
+ "model.layers.7.self_attn.k_norm.weight": {
3117
+ "shard": 5,
3118
+ "offset": 55688192,
3119
+ "size": 512,
3120
+ "shape": [
3121
+ 256
3122
+ ],
3123
+ "dtype": "BF16",
3124
+ "role": "norm"
3125
+ },
3126
+ "model.layers.7.self_attn.k_proj.weight": {
3127
+ "shard": 5,
3128
+ "offset": 55688704,
3129
+ "size": 110592,
3130
+ "shape": [
3131
+ 256,
3132
+ 640
3133
+ ],
3134
+ "dtype": "Q4_K_M",
3135
+ "role": "matmul",
3136
+ "layout": "row"
3137
+ },
3138
+ "model.layers.7.self_attn.o_proj.weight": {
3139
+ "shard": 5,
3140
+ "offset": 55799296,
3141
+ "size": 368640,
3142
+ "shape": [
3143
+ 640,
3144
+ 1024
3145
+ ],
3146
+ "dtype": "Q4_K_M",
3147
+ "role": "matmul",
3148
+ "layout": "row"
3149
+ },
3150
+ "model.layers.7.self_attn.q_norm.weight": {
3151
+ "shard": 5,
3152
+ "offset": 56167936,
3153
+ "size": 512,
3154
+ "shape": [
3155
+ 256
3156
+ ],
3157
+ "dtype": "BF16",
3158
+ "role": "norm"
3159
+ },
3160
+ "model.layers.7.self_attn.q_proj.weight": {
3161
+ "shard": 5,
3162
+ "offset": 56168448,
3163
+ "size": 442368,
3164
+ "shape": [
3165
+ 1024,
3166
+ 640
3167
+ ],
3168
+ "dtype": "Q4_K_M",
3169
+ "role": "matmul",
3170
+ "layout": "row"
3171
+ },
3172
+ "model.layers.7.self_attn.v_proj.weight": {
3173
+ "shard": 5,
3174
+ "offset": 56610816,
3175
+ "size": 110592,
3176
+ "shape": [
3177
+ 256,
3178
+ 640
3179
+ ],
3180
+ "dtype": "Q4_K_M",
3181
+ "role": "matmul",
3182
+ "layout": "row"
3183
+ },
3184
+ "model.layers.8.input_layernorm.weight": {
3185
+ "shard": 5,
3186
+ "offset": 56721408,
3187
+ "size": 1280,
3188
+ "shape": [
3189
+ 640
3190
+ ],
3191
+ "dtype": "BF16",
3192
+ "role": "norm"
3193
+ },
3194
+ "model.layers.8.mlp.down_proj.weight": {
3195
+ "shard": 5,
3196
+ "offset": 56722688,
3197
+ "size": 737280,
3198
+ "shape": [
3199
+ 640,
3200
+ 2048
3201
+ ],
3202
+ "dtype": "Q4_K_M",
3203
+ "role": "matmul",
3204
+ "layout": "row"
3205
+ },
3206
+ "model.layers.8.mlp.gate_proj.weight": {
3207
+ "shard": 5,
3208
+ "offset": 57459968,
3209
+ "size": 884736,
3210
+ "shape": [
3211
+ 2048,
3212
+ 640
3213
+ ],
3214
+ "dtype": "Q4_K_M",
3215
+ "role": "matmul",
3216
+ "layout": "row"
3217
+ },
3218
+ "model.layers.8.mlp.up_proj.weight": {
3219
+ "shard": 5,
3220
+ "offset": 58344704,
3221
+ "size": 884736,
3222
+ "shape": [
3223
+ 2048,
3224
+ 640
3225
+ ],
3226
+ "dtype": "Q4_K_M",
3227
+ "role": "matmul",
3228
+ "layout": "row"
3229
+ },
3230
+ "model.layers.8.post_attention_layernorm.weight": {
3231
+ "shard": 5,
3232
+ "offset": 59229440,
3233
+ "size": 1280,
3234
+ "shape": [
3235
+ 640
3236
+ ],
3237
+ "dtype": "BF16",
3238
+ "role": "norm"
3239
+ },
3240
+ "model.layers.8.post_feedforward_layernorm.weight": {
3241
+ "shard": 5,
3242
+ "offset": 59230720,
3243
+ "size": 1280,
3244
+ "shape": [
3245
+ 640
3246
+ ],
3247
+ "dtype": "BF16",
3248
+ "role": "norm"
3249
+ },
3250
+ "model.layers.8.pre_feedforward_layernorm.weight": {
3251
+ "shard": 5,
3252
+ "offset": 59232000,
3253
+ "size": 1280,
3254
+ "shape": [
3255
+ 640
3256
+ ],
3257
+ "dtype": "BF16",
3258
+ "role": "norm"
3259
+ },
3260
+ "model.layers.8.self_attn.k_norm.weight": {
3261
+ "shard": 5,
3262
+ "offset": 59233280,
3263
+ "size": 512,
3264
+ "shape": [
3265
+ 256
3266
+ ],
3267
+ "dtype": "BF16",
3268
+ "role": "norm"
3269
+ },
3270
+ "model.layers.8.self_attn.k_proj.weight": {
3271
+ "shard": 5,
3272
+ "offset": 59233792,
3273
+ "size": 110592,
3274
+ "shape": [
3275
+ 256,
3276
+ 640
3277
+ ],
3278
+ "dtype": "Q4_K_M",
3279
+ "role": "matmul",
3280
+ "layout": "row"
3281
+ },
3282
+ "model.layers.8.self_attn.o_proj.weight": {
3283
+ "shard": 5,
3284
+ "offset": 59344384,
3285
+ "size": 368640,
3286
+ "shape": [
3287
+ 640,
3288
+ 1024
3289
+ ],
3290
+ "dtype": "Q4_K_M",
3291
+ "role": "matmul",
3292
+ "layout": "row"
3293
+ },
3294
+ "model.layers.8.self_attn.q_norm.weight": {
3295
+ "shard": 5,
3296
+ "offset": 59713024,
3297
+ "size": 512,
3298
+ "shape": [
3299
+ 256
3300
+ ],
3301
+ "dtype": "BF16",
3302
+ "role": "norm"
3303
+ },
3304
+ "model.layers.8.self_attn.q_proj.weight": {
3305
+ "shard": 5,
3306
+ "offset": 59713536,
3307
+ "size": 442368,
3308
+ "shape": [
3309
+ 1024,
3310
+ 640
3311
+ ],
3312
+ "dtype": "Q4_K_M",
3313
+ "role": "matmul",
3314
+ "layout": "row"
3315
+ },
3316
+ "model.layers.8.self_attn.v_proj.weight": {
3317
+ "shard": 5,
3318
+ "offset": 60155904,
3319
+ "size": 110592,
3320
+ "shape": [
3321
+ 256,
3322
+ 640
3323
+ ],
3324
+ "dtype": "Q4_K_M",
3325
+ "role": "matmul",
3326
+ "layout": "row"
3327
+ },
3328
+ "model.layers.9.input_layernorm.weight": {
3329
+ "shard": 5,
3330
+ "offset": 60266496,
3331
+ "size": 1280,
3332
+ "shape": [
3333
+ 640
3334
+ ],
3335
+ "dtype": "BF16",
3336
+ "role": "norm"
3337
+ },
3338
+ "model.layers.9.mlp.down_proj.weight": {
3339
+ "shard": 5,
3340
+ "offset": 60267776,
3341
+ "size": 737280,
3342
+ "shape": [
3343
+ 640,
3344
+ 2048
3345
+ ],
3346
+ "dtype": "Q4_K_M",
3347
+ "role": "matmul",
3348
+ "layout": "row"
3349
+ },
3350
+ "model.layers.9.mlp.gate_proj.weight": {
3351
+ "shard": 5,
3352
+ "offset": 61005056,
3353
+ "size": 884736,
3354
+ "shape": [
3355
+ 2048,
3356
+ 640
3357
+ ],
3358
+ "dtype": "Q4_K_M",
3359
+ "role": "matmul",
3360
+ "layout": "row"
3361
+ },
3362
+ "model.layers.9.mlp.up_proj.weight": {
3363
+ "shard": 5,
3364
+ "offset": 61889792,
3365
+ "size": 884736,
3366
+ "shape": [
3367
+ 2048,
3368
+ 640
3369
+ ],
3370
+ "dtype": "Q4_K_M",
3371
+ "role": "matmul",
3372
+ "layout": "row"
3373
+ },
3374
+ "model.layers.9.post_attention_layernorm.weight": {
3375
+ "shard": 5,
3376
+ "offset": 62774528,
3377
+ "size": 1280,
3378
+ "shape": [
3379
+ 640
3380
+ ],
3381
+ "dtype": "BF16",
3382
+ "role": "norm"
3383
+ },
3384
+ "model.layers.9.post_feedforward_layernorm.weight": {
3385
+ "shard": 5,
3386
+ "offset": 62775808,
3387
+ "size": 1280,
3388
+ "shape": [
3389
+ 640
3390
+ ],
3391
+ "dtype": "BF16",
3392
+ "role": "norm"
3393
+ },
3394
+ "model.layers.9.pre_feedforward_layernorm.weight": {
3395
+ "shard": 5,
3396
+ "offset": 62777088,
3397
+ "size": 1280,
3398
+ "shape": [
3399
+ 640
3400
+ ],
3401
+ "dtype": "BF16",
3402
+ "role": "norm"
3403
+ },
3404
+ "model.layers.9.self_attn.k_norm.weight": {
3405
+ "shard": 5,
3406
+ "offset": 62778368,
3407
+ "size": 512,
3408
+ "shape": [
3409
+ 256
3410
+ ],
3411
+ "dtype": "BF16",
3412
+ "role": "norm"
3413
+ },
3414
+ "model.layers.9.self_attn.k_proj.weight": {
3415
+ "shard": 5,
3416
+ "offset": 62778880,
3417
+ "size": 110592,
3418
+ "shape": [
3419
+ 256,
3420
+ 640
3421
+ ],
3422
+ "dtype": "Q4_K_M",
3423
+ "role": "matmul",
3424
+ "layout": "row"
3425
+ },
3426
+ "model.layers.9.self_attn.o_proj.weight": {
3427
+ "shard": 5,
3428
+ "offset": 62889472,
3429
+ "size": 368640,
3430
+ "shape": [
3431
+ 640,
3432
+ 1024
3433
+ ],
3434
+ "dtype": "Q4_K_M",
3435
+ "role": "matmul",
3436
+ "layout": "row"
3437
+ },
3438
+ "model.layers.9.self_attn.q_norm.weight": {
3439
+ "shard": 5,
3440
+ "offset": 63258112,
3441
+ "size": 512,
3442
+ "shape": [
3443
+ 256
3444
+ ],
3445
+ "dtype": "BF16",
3446
+ "role": "norm"
3447
+ },
3448
+ "model.layers.9.self_attn.q_proj.weight": {
3449
+ "shard": 5,
3450
+ "offset": 63258624,
3451
+ "size": 442368,
3452
+ "shape": [
3453
+ 1024,
3454
+ 640
3455
+ ],
3456
+ "dtype": "Q4_K_M",
3457
+ "role": "matmul",
3458
+ "layout": "row"
3459
+ },
3460
+ "model.layers.9.self_attn.v_proj.weight": {
3461
+ "shard": 5,
3462
+ "offset": 63700992,
3463
+ "size": 110592,
3464
+ "shape": [
3465
+ 256,
3466
+ 640
3467
+ ],
3468
+ "dtype": "Q4_K_M",
3469
+ "role": "matmul",
3470
+ "layout": "row"
3471
+ },
3472
+ "model.norm.weight": {
3473
+ "shard": 5,
3474
+ "offset": 63811584,
3475
+ "size": 1280,
3476
+ "shape": [
3477
+ 640
3478
+ ],
3479
+ "dtype": "BF16",
3480
+ "role": "norm"
3481
+ }
3482
+ },
3483
+ "totalSize": 399357184,
3484
+ "hashAlgorithm": "blake3",
3485
+ "eos_token_id": [
3486
+ 1,
3487
+ 50
3488
+ ],
3489
+ "metadata": {
3490
+ "source": "convert-core",
3491
+ "convertedAt": "2026-02-26T17:04:12.403Z",
3492
+ "hasTokenizer": true,
3493
+ "manifestRefresh": {
3494
+ "at": "2026-02-26T20:36:34.427Z",
3495
+ "config": "gemma-3-270m-it-wq4k-ef16-hf16-f32.json"
3496
+ }
3497
+ },
3498
+ "tokenizer": {
3499
+ "type": "bundled",
3500
+ "vocabSize": 3119,
3501
+ "file": "tokenizer.json"
3502
+ }
3503
+ }
models/gemma-3-270m-it-wq4k-ef16/shard_00000.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40d8ab4be58717665fef7a3bbffb946f2a406db1318dce18c20ad36a98e781ba
3
+ size 67108864
models/gemma-3-270m-it-wq4k-ef16/shard_00001.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4fe25ceb18b30aef9abe5cf59f7a21c9bc2cbfa578c74c0e4e39b0b5cae02388
3
+ size 67108864
models/gemma-3-270m-it-wq4k-ef16/shard_00002.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f28852ac459865222c8fac27e74ca4fa77729c47302b5e045c5cc27413912ff
3
+ size 67108864
models/gemma-3-270m-it-wq4k-ef16/shard_00003.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48641dd3c83bfd8b96408e4d2cbaa6f7d39a64e883b2226f8646bf95f13472b2
3
+ size 67108864
models/gemma-3-270m-it-wq4k-ef16/shard_00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2aff6d503de3e0d28905ce96691163288ac3b008dcc9591f4f38c98c73263a7c
3
+ size 67108864
models/gemma-3-270m-it-wq4k-ef16/shard_00005.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:590071d94ec289c2a08173c79c4280528cd6dd1ba08d433de6050b4b518b1f7e
3
+ size 63812864
models/gemma-3-270m-it-wq4k-ef16/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb9a1540f10b955cf49c205984c4ed9cabe5454280ec50fb4f2fae5904af0f2a
3
+ size 14386512
models/gemma-3-270m-it-wq4k-ef16/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa009fcbc3589a9904d30d04834094fea4653c2ac6d2de2cd1262d4f7a50ceb3
3
+ size 4689144