Run 11. Outer Step 236. Inner Step 9.
Browse files- config.json +51 -51
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,26 +8,26 @@
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
-
"104": "
|
| 12 |
-
"105": "
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "SUCCESS",
|
| 15 |
"108": "SUCCESS",
|
| 16 |
-
"109": "
|
| 17 |
-
"11": "
|
| 18 |
-
"110": "
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
-
"116": "
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
-
"119": "
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
-
"120": "
|
| 30 |
-
"121": "
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
|
@@ -39,24 +39,24 @@
|
|
| 39 |
"13": "SUCCESS",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
-
"132": "
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
-
"134": "
|
| 45 |
"135": "SUCCESS",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
-
"138": "
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "SUCCESS",
|
| 59 |
-
"148": "
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
|
@@ -74,7 +74,7 @@
|
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "SUCCESS",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
@@ -83,28 +83,28 @@
|
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
-
"172": "
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
-
"178": "
|
| 93 |
-
"179": "
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
-
"181": "
|
| 97 |
"182": "SUCCESS",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
"186": "SUCCESS",
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
-
"191": "
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
|
@@ -116,15 +116,15 @@
|
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "SUCCESS",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
-
"201": "
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
-
"204": "
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
-
"209": "
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
|
@@ -133,14 +133,14 @@
|
|
| 133 |
"214": "SUCCESS",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "SUCCESS",
|
|
@@ -154,15 +154,15 @@
|
|
| 154 |
"233": "SUCCESS",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
-
"236": "
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
-
"24": "
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "SUCCESS",
|
| 164 |
-
"242": "
|
| 165 |
-
"243": "
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "NOT_ALIVE",
|
|
@@ -185,13 +185,13 @@
|
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
-
"37": "
|
| 192 |
-
"38": "
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
-
"4": "
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
|
@@ -200,14 +200,14 @@
|
|
| 200 |
"45": "NON_PARTICIPATING",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
-
"48": "
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
-
"54": "
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
|
@@ -220,9 +220,9 @@
|
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
-
"66": "
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
-
"68": "
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
|
@@ -235,27 +235,27 @@
|
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
-
"8": "
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
-
"82": "
|
| 242 |
-
"83": "
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
-
"9": "
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
-
"91": "
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
-
"95": "
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
-
"98": "
|
| 259 |
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -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",
|
|
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
+
"104": "NOT_ALIVE",
|
| 12 |
+
"105": "NOT_ALIVE",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "SUCCESS",
|
| 15 |
"108": "SUCCESS",
|
| 16 |
+
"109": "NOT_ALIVE",
|
| 17 |
+
"11": "NON_PARTICIPATING",
|
| 18 |
+
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
+
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
+
"119": "SUCCESS",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
+
"120": "SUCCESS",
|
| 30 |
+
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
|
|
|
| 39 |
"13": "SUCCESS",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
+
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
+
"134": "SUCCESS",
|
| 45 |
"135": "SUCCESS",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
+
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "SUCCESS",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
|
|
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "SUCCESS",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
|
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
+
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
+
"178": "NON_PARTICIPATING",
|
| 93 |
+
"179": "NOT_ALIVE",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
+
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "SUCCESS",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "NOT_ALIVE",
|
| 101 |
"186": "SUCCESS",
|
| 102 |
"187": "SUCCESS",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
+
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
|
|
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "SUCCESS",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
+
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
+
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
+
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
+
"209": "SUCCESS",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
|
|
|
| 133 |
"214": "SUCCESS",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
+
"217": "NOT_ALIVE",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "SUCCESS",
|
|
|
|
| 154 |
"233": "SUCCESS",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
+
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
+
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "SUCCESS",
|
| 164 |
+
"242": "NON_PARTICIPATING",
|
| 165 |
+
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "NOT_ALIVE",
|
|
|
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "NOT_ALIVE",
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
+
"37": "SUCCESS",
|
| 192 |
+
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
+
"4": "SUCCESS",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
|
|
|
| 200 |
"45": "NON_PARTICIPATING",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
+
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
+
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
|
|
|
| 220 |
"63": "NOT_ALIVE",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
+
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
+
"68": "SUCCESS",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
|
|
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
+
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "SUCCESS",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
+
"82": "NOT_ALIVE",
|
| 242 |
+
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
"86": "NOT_ALIVE",
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
+
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
+
"91": "NOT_ALIVE",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
+
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
+
"98": "SUCCESS",
|
| 259 |
"99": "SUCCESS"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6673111,
|
| 268 |
+
6673140
|
| 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": 6673078,
|
| 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:f6e4e3b940cd0253962ae9ea7e54ceb61786f154637ba9275749f9ba43a59267
|
| 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 sha256:240dece264699e29f4247d824cf4e7cfb9913894dd9b9bdcd180b835ccf36513
|
| 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 sha256:2c84b4368c79b9aa7dd3deb0d65c704b1d67048f9462af9c210e3bcf6ce5e304
|
| 3 |
size 4400216536
|