Run 6. Outer Step 78. Inner Step 5.
Browse files- config.json +51 -51
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -6,12 +6,12 @@
|
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
-
"102": "
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
-
"107": "
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
|
@@ -20,10 +20,10 @@
|
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
-
"118": "
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
|
@@ -33,59 +33,59 @@
|
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
-
"127": "
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
-
"139": "
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
-
"154": "
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
-
"158": "
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
-
"165": "
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
-
"168": "
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
-
"173": "
|
| 88 |
-
"174": "
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
@@ -93,24 +93,24 @@
|
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
-
"181": "
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
-
"185": "
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
-
"197": "
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
|
@@ -118,15 +118,15 @@
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "SUCCESS",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
-
"203": "
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
-
"209": "
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
"213": "NOT_ALIVE",
|
|
@@ -135,23 +135,23 @@
|
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
-
"225": "
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
-
"228": "
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
-
"233": "
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
@@ -163,7 +163,7 @@
|
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
-
"244": "
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
|
@@ -172,15 +172,15 @@
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
-
"253": "
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
-
"3": "
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
|
@@ -188,7 +188,7 @@
|
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
-
"37": "
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
|
@@ -219,21 +219,21 @@
|
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
-
"65": "
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
-
"70": "
|
| 229 |
-
"71": "
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
-
"76": "
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
-
"78": "
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
|
@@ -249,8 +249,8 @@
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
-
"92": "
|
| 253 |
-
"93": "
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
|
@@ -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",
|
|
|
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
+
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
+
"107": "NOT_ALIVE",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
|
|
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
+
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
|
|
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
+
"127": "NOT_ALIVE",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "SUCCESS",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
+
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
+
"154": "SUCCESS",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
+
"158": "NON_PARTICIPATING",
|
| 71 |
+
"159": "NOT_ALIVE",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
+
"168": "NOT_ALIVE",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
+
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
|
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
+
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
+
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
+
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
+
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
|
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "SUCCESS",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
+
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
+
"209": "SUCCESS",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "NOT_ALIVE",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
"213": "NOT_ALIVE",
|
|
|
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
+
"219": "NON_PARTICIPATING",
|
| 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": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NOT_ALIVE",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
|
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
+
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
|
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "NON_PARTICIPATING",
|
| 176 |
+
"253": "NOT_ALIVE",
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
+
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
|
|
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
+
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
|
|
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
+
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
+
"70": "NON_PARTICIPATING",
|
| 229 |
+
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
+
"76": "NOT_ALIVE",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
+
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
|
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "NON_PARTICIPATING",
|
| 253 |
+
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6183538,
|
| 268 |
+
6183557
|
| 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": 5,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6183354,
|
| 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 sha256:43f44b18ea43d5bf0d2493cd5f180bd0d765a581fcb22cc1fc1d22734b3f2cff
|
| 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:ea9ea818c18e2078212922592bf171a5782d42900a595650eecd14c287daa2e8
|
| 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 sha256:5734b5493ce143026d60e68d1ae6214b0ec86d4874a3ac12c1808a30daadd3fc
|
| 3 |
size 4400216536
|