Run 11. Outer Step 45. Inner Step 9.
Browse files- config.json +97 -97
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -1,59 +1,59 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
-
"102": "
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
-
"105": "
|
| 13 |
-
"106": "
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
-
"109": "
|
| 17 |
-
"11": "
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
-
"111": "
|
| 20 |
-
"112": "
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
-
"115": "
|
| 24 |
-
"116": "
|
| 25 |
"117": "SUCCESS",
|
| 26 |
-
"118": "
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
-
"120": "
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
-
"123": "
|
| 33 |
"124": "SUCCESS",
|
| 34 |
-
"125": "
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
-
"128": "
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
-
"13": "
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
-
"14": "
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
-
"142": "
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
|
@@ -62,103 +62,103 @@
|
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
-
"156": "
|
| 69 |
-
"157": "
|
| 70 |
-
"158": "
|
| 71 |
-
"159": "
|
| 72 |
-
"16": "
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
-
"168": "
|
| 82 |
-
"169": "
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
-
"174": "
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
-
"18": "
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
-
"184": "
|
| 100 |
-
"185": "
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
-
"188": "
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
-
"192": "
|
| 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": "
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
-
"203": "
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
-
"205": "
|
| 124 |
-
"206": "
|
| 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",
|
| 140 |
"220": "SUCCESS",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
-
"222": "
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
-
"226": "
|
| 147 |
-
"227": "
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
-
"232": "
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "FAIL",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
-
"237": "
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
-
"24": "
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "NON_PARTICIPATING",
|
|
@@ -172,91 +172,91 @@
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NOT_ALIVE",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
-
"252": "
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
-
"26": "
|
| 180 |
-
"27": "
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
-
"3": "
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
-
"31": "
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
"35": "FAIL",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
-
"38": "
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
-
"41": "
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
-
"44": "
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
-
"46": "
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
-
"48": "
|
| 204 |
-
"49": "
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
-
"52": "
|
| 209 |
-
"53": "
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
-
"55": "
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
-
"6": "
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
-
"62": "
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
-
"64": "
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
-
"67": "
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
-
"69": "
|
| 227 |
-
"7": "
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
-
"75": "
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
-
"77": "
|
| 236 |
-
"78": "
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "NON_PARTICIPATING",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
-
"83": "
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
-
"87": "
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "FAIL",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
"91": "NON_PARTICIPATING",
|
| 252 |
"92": "FAIL",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
-
"95": "
|
| 256 |
-
"96": "
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
"98": "NOT_ALIVE",
|
| 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",
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "dstrbtd/llama-1b",
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
+
"102": "FAIL",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
+
"105": "SUCCESS",
|
| 13 |
+
"106": "SUCCESS",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
+
"109": "SUCCESS",
|
| 17 |
+
"11": "NON_PARTICIPATING",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
+
"111": "NOT_ALIVE",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
+
"115": "SUCCESS",
|
| 24 |
+
"116": "SUCCESS",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
+
"118": "SUCCESS",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
+
"120": "NOT_ALIVE",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
+
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
+
"125": "SUCCESS",
|
| 35 |
+
"126": "NOT_ALIVE",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
+
"128": "FAIL",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
+
"13": "SUCCESS",
|
| 40 |
+
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
+
"135": "SUCCESS",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "SUCCESS",
|
| 50 |
+
"14": "SUCCESS",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
+
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
+
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
|
|
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "FAIL",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
+
"156": "NOT_ALIVE",
|
| 69 |
+
"157": "SUCCESS",
|
| 70 |
+
"158": "SUCCESS",
|
| 71 |
+
"159": "SUCCESS",
|
| 72 |
+
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "NOT_ALIVE",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 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": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
+
"184": "FAIL",
|
| 100 |
+
"185": "FAIL",
|
| 101 |
+
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
+
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
+
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "NON_PARTICIPATING",
|
| 114 |
+
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
+
"20": "NON_PARTICIPATING",
|
| 118 |
+
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
+
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
+
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
+
"21": "SUCCESS",
|
| 129 |
"210": "SUCCESS",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
+
"212": "FAIL",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "NOT_ALIVE",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "SUCCESS",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
+
"222": "SUCCESS",
|
| 143 |
+
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
+
"226": "NOT_ALIVE",
|
| 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": "FAIL",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
+
"237": "SUCCESS",
|
| 159 |
+
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
+
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NOT_ALIVE",
|
| 164 |
"242": "NON_PARTICIPATING",
|
|
|
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NOT_ALIVE",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
+
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
+
"26": "FAIL",
|
| 180 |
+
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
+
"3": "SUCCESS",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
+
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "FAIL",
|
| 189 |
"35": "FAIL",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "NON_PARTICIPATING",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
+
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
+
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
+
"46": "NOT_ALIVE",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
+
"48": "FAIL",
|
| 204 |
+
"49": "SUCCESS",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
+
"52": "NON_PARTICIPATING",
|
| 209 |
+
"53": "FAIL",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
+
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
+
"57": "FAIL",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
+
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "SUCCESS",
|
| 219 |
+
"62": "SUCCESS",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
+
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
+
"67": "NON_PARTICIPATING",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
+
"69": "SUCCESS",
|
| 227 |
+
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
+
"73": "SUCCESS",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
+
"75": "NOT_ALIVE",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "SUCCESS",
|
| 236 |
+
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "NON_PARTICIPATING",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
+
"83": "FAIL",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "SUCCESS",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
+
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "FAIL",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NON_PARTICIPATING",
|
| 252 |
"92": "FAIL",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
+
"95": "NON_PARTICIPATING",
|
| 256 |
+
"96": "SUCCESS",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
"98": "NOT_ALIVE",
|
| 259 |
+
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6567362,
|
| 268 |
+
6567390
|
| 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": 9,
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
+
"last_allreduce_block": 6567295,
|
| 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 sha256:8f7bd738e0f4003c27be60fd9d3b51665bd0f11067bc53c23e81d25de976397f
|
| 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:e29de4a608be02c30680da1dc4d44b63e47d11a0413459699207419b363d06d4
|
| 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:a02acd5beac2fb984d48cf5883d399539e55dca20bc36ea1328df6bc5d17c98a
|
| 3 |
size 4400216536
|