Run 6. Outer Step 51. Inner Step 5.
Browse files- config.json +38 -38
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -27,7 +27,7 @@
|
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
"124": "NON_PARTICIPATING",
|
|
@@ -40,48 +40,48 @@
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
-
"137": "
|
| 48 |
-
"138": "
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
-
"151": "
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "SUCCESS",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
-
"158": "
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
-
"163": "
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
-
"169": "
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
-
"170": "
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NON_PARTICIPATING",
|
|
@@ -89,7 +89,7 @@
|
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
-
"178": "
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
|
@@ -98,14 +98,14 @@
|
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
-
"186": "
|
| 102 |
-
"187": "
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
-
"189": "
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
-
"192": "
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
|
@@ -116,7 +116,7 @@
|
|
| 116 |
"2": "SUCCESS",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
-
"201": "
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
|
@@ -128,27 +128,27 @@
|
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
-
"212": "
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
-
"224": "
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
|
@@ -161,7 +161,7 @@
|
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
-
"242": "
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NOT_ALIVE",
|
|
@@ -185,7 +185,7 @@
|
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
-
"34": "
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
|
@@ -215,30 +215,30 @@
|
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
-
"66": "
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
-
"70": "
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
-
"72": "
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
-
"76": "
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
-
"82": "
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
|
@@ -252,10 +252,10 @@
|
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
-
"95": "
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
-
"98": "
|
| 259 |
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,17 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
-
"inner_step":
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
-
"last_allreduce_block":
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
|
|
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
+
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
"124": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NOT_ALIVE",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
+
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "NOT_ALIVE",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
+
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "SUCCESS",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
+
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
+
"166": "FAIL",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
+
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NON_PARTICIPATING",
|
|
|
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
+
"178": "NOT_ALIVE",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
|
|
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
+
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
+
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
|
|
|
| 116 |
"2": "SUCCESS",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
+
"201": "FAIL",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
|
|
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
+
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
+
"219": "FAIL",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
+
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
|
|
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
+
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NOT_ALIVE",
|
|
|
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
+
"34": "NOT_ALIVE",
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
|
|
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
+
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
+
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
+
"72": "NOT_ALIVE",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
+
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
+
"82": "NOT_ALIVE",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
|
|
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
+
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
+
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6163711,
|
| 268 |
+
6163755
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
+
"inner_step": 4,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6163627,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:48e13af59731d48b840fb9ed5594bba4495b869c224f6dfe37ef31f84723634c
|
| 3 |
size 88389626
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c84cba7b6afdb6a37ea4ba7aedddb4cd30fd87f0fdabe9e4f6ecbd237d96868d
|
| 3 |
size 8800564076
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4400216536
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34bb93873956e3b0e8a08550ee86672831b4e6508277d561ca1262e00288c148
|
| 3 |
size 4400216536
|