Upload Phi3ForCausalLM
Browse files- config.json +8 -101
config.json
CHANGED
|
@@ -22,114 +22,21 @@
|
|
| 22 |
"rms_norm_eps": 1e-05,
|
| 23 |
"rope_scaling": {
|
| 24 |
"long_factor": [
|
| 25 |
-
1
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
1.5699999332427979,
|
| 30 |
-
1.7999999523162842,
|
| 31 |
-
2.129999876022339,
|
| 32 |
-
2.129999876022339,
|
| 33 |
-
3.009999990463257,
|
| 34 |
-
5.910000324249268,
|
| 35 |
-
6.950000286102295,
|
| 36 |
-
9.070000648498535,
|
| 37 |
-
9.930000305175781,
|
| 38 |
-
10.710000038146973,
|
| 39 |
-
11.130000114440918,
|
| 40 |
-
14.609999656677246,
|
| 41 |
-
15.409998893737793,
|
| 42 |
-
19.809999465942383,
|
| 43 |
-
37.279998779296875,
|
| 44 |
-
38.279998779296875,
|
| 45 |
-
38.599998474121094,
|
| 46 |
-
40.12000274658203,
|
| 47 |
-
46.20000457763672,
|
| 48 |
-
50.940006256103516,
|
| 49 |
-
53.66000747680664,
|
| 50 |
-
54.9373893737793,
|
| 51 |
-
56.89738845825195,
|
| 52 |
-
57.28738784790039,
|
| 53 |
-
59.98738479614258,
|
| 54 |
-
60.86738586425781,
|
| 55 |
-
60.887386322021484,
|
| 56 |
-
61.71739196777344,
|
| 57 |
-
62.91739273071289,
|
| 58 |
-
62.957393646240234,
|
| 59 |
-
63.41739273071289,
|
| 60 |
-
63.8173942565918,
|
| 61 |
-
63.83739471435547,
|
| 62 |
-
63.897396087646484,
|
| 63 |
-
63.93739700317383,
|
| 64 |
-
64.06739807128906,
|
| 65 |
-
64.11434936523438,
|
| 66 |
-
64.12435150146484,
|
| 67 |
-
64.15435028076172,
|
| 68 |
-
64.19435119628906,
|
| 69 |
-
64.24435424804688,
|
| 70 |
-
64.57435607910156,
|
| 71 |
-
64.69000244140625,
|
| 72 |
-
64.76000213623047
|
| 73 |
],
|
| 74 |
"short_factor": [
|
| 75 |
-
1
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
1.3500000000000003,
|
| 80 |
-
1.3500000000000003,
|
| 81 |
-
1.4000000000000004,
|
| 82 |
-
1.5500000000000005,
|
| 83 |
-
2.000000000000001,
|
| 84 |
-
2.000000000000001,
|
| 85 |
-
2.000000000000001,
|
| 86 |
-
2.000000000000001,
|
| 87 |
-
2.000000000000001,
|
| 88 |
-
2.000000000000001,
|
| 89 |
-
2.000000000000001,
|
| 90 |
-
2.000000000000001,
|
| 91 |
-
2.000000000000001,
|
| 92 |
-
2.000000000000001,
|
| 93 |
-
2.000000000000001,
|
| 94 |
-
2.000000000000001,
|
| 95 |
-
2.000000000000001,
|
| 96 |
-
2.000000000000001,
|
| 97 |
-
2.000000000000001,
|
| 98 |
-
2.000000000000001,
|
| 99 |
-
2.000000000000001,
|
| 100 |
-
2.0500000000000007,
|
| 101 |
-
2.0500000000000007,
|
| 102 |
-
2.0500000000000007,
|
| 103 |
-
2.0500000000000007,
|
| 104 |
-
2.0500000000000007,
|
| 105 |
-
2.0500000000000007,
|
| 106 |
-
2.1000000000000005,
|
| 107 |
-
2.1000000000000005,
|
| 108 |
-
2.1500000000000004,
|
| 109 |
-
2.25,
|
| 110 |
-
2.25,
|
| 111 |
-
2.25,
|
| 112 |
-
2.25,
|
| 113 |
-
2.25,
|
| 114 |
-
2.3999999999999995,
|
| 115 |
-
2.4499999999999993,
|
| 116 |
-
2.499999999999999,
|
| 117 |
-
2.6999999999999984,
|
| 118 |
-
2.6999999999999984,
|
| 119 |
-
2.7499999999999982,
|
| 120 |
-
2.799999999999998,
|
| 121 |
-
2.8999999999999977,
|
| 122 |
-
3.049999999999997
|
| 123 |
],
|
| 124 |
"type": "longrope"
|
| 125 |
},
|
| 126 |
"rope_theta": 10000.0,
|
| 127 |
"sliding_window": 262144,
|
| 128 |
-
"resid_pdrop": 0.0,
|
| 129 |
-
"rms_norm_eps": 1e-05,
|
| 130 |
-
"rope_scaling": null,
|
| 131 |
-
"rope_theta": 10000.0,
|
| 132 |
-
"sliding_window": 2047,
|
| 133 |
"tie_word_embeddings": false,
|
| 134 |
"torch_dtype": "float32",
|
| 135 |
"transformers_version": "4.54.0.dev0",
|
|
|
|
| 22 |
"rms_norm_eps": 1e-05,
|
| 23 |
"rope_scaling": {
|
| 24 |
"long_factor": [
|
| 25 |
+
1,
|
| 26 |
+
2,
|
| 27 |
+
8,
|
| 28 |
+
64
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
],
|
| 30 |
"short_factor": [
|
| 31 |
+
1,
|
| 32 |
+
2,
|
| 33 |
+
3,
|
| 34 |
+
4
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
],
|
| 36 |
"type": "longrope"
|
| 37 |
},
|
| 38 |
"rope_theta": 10000.0,
|
| 39 |
"sliding_window": 262144,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
"tie_word_embeddings": false,
|
| 41 |
"torch_dtype": "float32",
|
| 42 |
"transformers_version": "4.54.0.dev0",
|