Run 8. Outer Step 42. Inner Step 4.
Browse files- config.json +55 -56
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -15,13 +15,13 @@
|
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
-
"113": "
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
-
"116": "
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "SUCCESS",
|
| 27 |
"119": "NOT_ALIVE",
|
|
@@ -29,7 +29,7 @@
|
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "FAIL",
|
| 32 |
-
"123": "
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
@@ -40,10 +40,10 @@
|
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
|
@@ -59,12 +59,12 @@
|
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
@@ -72,13 +72,13 @@
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "FAIL",
|
| 75 |
-
"162": "
|
| 76 |
-
"163": "
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
-
"168": "
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
|
@@ -86,42 +86,42 @@
|
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "SUCCESS",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
-
"179": "
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
-
"180": "
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
-
"185": "
|
| 101 |
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "FAIL",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
-
"190": "
|
| 107 |
"191": "FAIL",
|
| 108 |
-
"192": "
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
"203": "FAIL",
|
| 122 |
-
"204": "
|
| 123 |
"205": "SUCCESS",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NON_PARTICIPATING",
|
|
@@ -134,22 +134,22 @@
|
|
| 134 |
"215": "FAIL",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
-
"218": "
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
-
"222": "
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
-
"224": "
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
-
"228": "
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
-
"231": "
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "SUCCESS",
|
|
@@ -158,7 +158,7 @@
|
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "FAIL",
|
| 161 |
-
"24": "
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "SUCCESS",
|
| 164 |
"242": "NON_PARTICIPATING",
|
|
@@ -168,21 +168,21 @@
|
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "FAIL",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "FAIL",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
"253": "SUCCESS",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "SUCCESS",
|
| 181 |
-
"28": "
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "FAIL",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
-
"31": "
|
| 186 |
"32": "FAIL",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
@@ -190,38 +190,38 @@
|
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
-
"41": "
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
-
"43": "
|
| 199 |
"44": "SUCCESS",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
-
"53": "
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
-
"6": "
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
-
"66": "
|
| 224 |
-
"67": "
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
|
@@ -236,11 +236,11 @@
|
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
-
"84": "
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "NOT_ALIVE",
|
|
@@ -248,14 +248,14 @@
|
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
-
"91": "
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
-
"98": "
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -264,18 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
6368646
|
| 270 |
],
|
| 271 |
"bos_token_id": 1,
|
| 272 |
"eos_token_id": 2,
|
| 273 |
"hidden_act": "silu",
|
| 274 |
"hidden_size": 2048,
|
| 275 |
"initializer_range": 0.02,
|
| 276 |
-
"inner_step":
|
| 277 |
"intermediate_size": 5632,
|
| 278 |
-
"last_allreduce_block":
|
| 279 |
"max_position_embeddings": 2048,
|
| 280 |
"mlp_bias": false,
|
| 281 |
"model_type": "llama",
|
|
|
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
+
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
+
"116": "NOT_ALIVE",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "SUCCESS",
|
| 27 |
"119": "NOT_ALIVE",
|
|
|
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "FAIL",
|
| 32 |
+
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
+
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
|
|
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
+
"150": "SUCCESS",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
+
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "FAIL",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
+
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
|
|
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "SUCCESS",
|
| 88 |
"174": "SUCCESS",
|
| 89 |
+
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
+
"179": "SUCCESS",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
+
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NOT_ALIVE",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
+
"185": "SUCCESS",
|
| 101 |
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "FAIL",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
+
"190": "NOT_ALIVE",
|
| 107 |
"191": "FAIL",
|
| 108 |
+
"192": "SUCCESS",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "FAIL",
|
| 121 |
"203": "FAIL",
|
| 122 |
+
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "SUCCESS",
|
| 124 |
+
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NON_PARTICIPATING",
|
|
|
|
| 134 |
"215": "FAIL",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
+
"218": "SUCCESS",
|
| 138 |
+
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "SUCCESS",
|
| 142 |
+
"222": "SUCCESS",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
+
"224": "NOT_ALIVE",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
+
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "NON_PARTICIPATING",
|
| 152 |
+
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "SUCCESS",
|
|
|
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "FAIL",
|
| 161 |
+
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "SUCCESS",
|
| 164 |
"242": "NON_PARTICIPATING",
|
|
|
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "FAIL",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
+
"249": "SUCCESS",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "FAIL",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "SUCCESS",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
+
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "SUCCESS",
|
| 181 |
+
"28": "SUCCESS",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "FAIL",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
+
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "FAIL",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
|
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
+
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
+
"43": "NOT_ALIVE",
|
| 199 |
"44": "SUCCESS",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
+
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
+
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
+
"6": "FAIL",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
+
"66": "NOT_ALIVE",
|
| 224 |
+
"67": "NOT_ALIVE",
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
|
|
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
+
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "NOT_ALIVE",
|
|
|
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
+
"98": "FAIL",
|
| 259 |
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6368858,
|
| 268 |
+
6368871
|
|
|
|
| 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": 6368820,
|
| 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 sha256:4b8bb115ffa21f39371fdc4764a37a4274490df8a4c8ba93deb552c5bd5bd09d
|
| 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 4924926906
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:07623eab26e23d78dcd913302090a3acf434f46b5aec4c804bb07a413d82eafa
|
| 3 |
size 4924926906
|
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 sha256:4032e74db5c33846ed082e4038b19be72a51ac7abb46cfaa583f5424651f20fa
|
| 3 |
size 4400216536
|