Run 11. Outer Step 42. Inner Step 9.
Browse files- config.json +57 -57
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -6,9 +6,9 @@
|
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
-
"102": "
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
-
"104": "
|
| 12 |
"105": "SUCCESS",
|
| 13 |
"106": "SUCCESS",
|
| 14 |
"107": "NOT_ALIVE",
|
|
@@ -22,7 +22,7 @@
|
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "SUCCESS",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
-
"117": "
|
| 26 |
"118": "SUCCESS",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
|
@@ -32,47 +32,47 @@
|
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "SUCCESS",
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
-
"128": "
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
-
"13": "
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
-
"132": "
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
-
"134": "
|
| 45 |
-
"135": "
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
-
"138": "
|
| 49 |
"139": "SUCCESS",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
-
"142": "
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
-
"158": "
|
| 71 |
"159": "SUCCESS",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
|
@@ -83,15 +83,15 @@
|
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
-
"172": "
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
-
"174": "
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
-
"18": "
|
| 95 |
"180": "SUCCESS",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
|
@@ -102,19 +102,19 @@
|
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
-
"19": "
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
-
"198": "
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
-
"20": "
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
|
@@ -125,15 +125,15 @@
|
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
-
"21": "
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
-
"212": "
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
|
@@ -143,14 +143,14 @@
|
|
| 143 |
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
-
"226": "
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
-
"229": "
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
-
"232": "
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NOT_ALIVE",
|
|
@@ -162,15 +162,15 @@
|
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
-
"243": "
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
-
"246": "
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
-
"248": "
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
|
@@ -181,12 +181,12 @@
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
-
"30": "
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
|
@@ -198,29 +198,29 @@
|
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
-
"46": "
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
-
"49": "
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
-
"53": "
|
| 210 |
-
"54": "
|
| 211 |
"55": "SUCCESS",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
-
"60": "
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
-
"63": "
|
| 221 |
-
"64": "
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
-
"66": "
|
| 224 |
"67": "NON_PARTICIPATING",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "SUCCESS",
|
|
@@ -229,20 +229,20 @@
|
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
-
"74": "
|
| 233 |
-
"75": "
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
-
"77": "
|
| 236 |
-
"78": "
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
-
"83": "
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "SUCCESS",
|
| 245 |
-
"86": "
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
|
@@ -255,8 +255,8 @@
|
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
-
"98": "
|
| 259 |
-
"99": "
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
@@ -264,8 +264,8 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"dtype": "float32",
|
|
@@ -274,9 +274,9 @@
|
|
| 274 |
"hidden_act": "silu",
|
| 275 |
"hidden_size": 2048,
|
| 276 |
"initializer_range": 0.02,
|
| 277 |
-
"inner_step":
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
-
"last_allreduce_block":
|
| 280 |
"max_position_embeddings": 2048,
|
| 281 |
"mlp_bias": false,
|
| 282 |
"model_type": "llama",
|
|
|
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
+
"102": "NOT_ALIVE",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
+
"104": "NOT_ALIVE",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
"106": "SUCCESS",
|
| 14 |
"107": "NOT_ALIVE",
|
|
|
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "SUCCESS",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
+
"117": "NOT_ALIVE",
|
| 26 |
"118": "SUCCESS",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
|
|
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "SUCCESS",
|
| 35 |
+
"126": "NOT_ALIVE",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
+
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
+
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
+
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
+
"134": "NOT_ALIVE",
|
| 45 |
+
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
+
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
+
"149": "SUCCESS",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
+
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
+
"158": "NON_PARTICIPATING",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
|
|
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
+
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
+
"18": "NOT_ALIVE",
|
| 95 |
"180": "SUCCESS",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
|
|
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
+
"19": "SUCCESS",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
+
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "SUCCESS",
|
| 114 |
+
"198": "SUCCESS",
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
+
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
|
|
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
+
"21": "NOT_ALIVE",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
+
"212": "NOT_ALIVE",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "SUCCESS",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
|
|
|
| 143 |
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
+
"226": "SUCCESS",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
+
"229": "NOT_ALIVE",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
+
"232": "NON_PARTICIPATING",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NOT_ALIVE",
|
|
|
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
+
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
+
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
+
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "SUCCESS",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
|
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
+
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
+
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
|
|
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
+
"46": "SUCCESS",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
+
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
+
"53": "NOT_ALIVE",
|
| 210 |
+
"54": "NOT_ALIVE",
|
| 211 |
"55": "SUCCESS",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
"58": "SUCCESS",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
+
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
+
"63": "NOT_ALIVE",
|
| 221 |
+
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
+
"66": "NOT_ALIVE",
|
| 224 |
"67": "NON_PARTICIPATING",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "SUCCESS",
|
|
|
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
+
"74": "NON_PARTICIPATING",
|
| 233 |
+
"75": "SUCCESS",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NOT_ALIVE",
|
| 236 |
+
"78": "SUCCESS",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
+
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "SUCCESS",
|
| 245 |
+
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
|
|
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
+
"98": "SUCCESS",
|
| 259 |
+
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6565161,
|
| 268 |
+
6565188
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"dtype": "float32",
|
|
|
|
| 274 |
"hidden_act": "silu",
|
| 275 |
"hidden_size": 2048,
|
| 276 |
"initializer_range": 0.02,
|
| 277 |
+
"inner_step": 8,
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
+
"last_allreduce_block": 6565137,
|
| 280 |
"max_position_embeddings": 2048,
|
| 281 |
"mlp_bias": false,
|
| 282 |
"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 sha256:0a284c9f0d620b2e9cd9221a0eada95d4a7d060255617655d1459e619aa3a748
|
| 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 sha256:660422f7276fbecfcb735deeabbded4bf32c64badd408f5bae1d73237f48e9e2
|
| 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 sha256:9bd48d861085ea08b334adfe01ae1fa09d28443fd9381bc29c15506b85ba7f59
|
| 3 |
size 4400216536
|