Run 5. Outer Step 126. Inner Step 1.
Browse files- config.json +54 -54
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -17,12 +17,12 @@
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
-
"113": "
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
|
@@ -35,57 +35,57 @@
|
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
-
"129": "
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
-
"138": "
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
-
"142": "
|
| 54 |
-
"143": "
|
| 55 |
-
"144": "
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
-
"147": "
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
-
"15": "
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
-
"168": "
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
-
"174": "
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
@@ -108,7 +108,7 @@
|
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
-
"195": "
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NON_PARTICIPATING",
|
|
@@ -116,28 +116,28 @@
|
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
-
"201": "
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
-
"208": "
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
-
"213": "
|
| 133 |
-
"214": "
|
| 134 |
-
"215": "
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
-
"220": "
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
|
@@ -155,8 +155,8 @@
|
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
-
"237": "
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
|
@@ -166,15 +166,15 @@
|
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
|
@@ -183,7 +183,7 @@
|
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
@@ -199,25 +199,25 @@
|
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
-
"47": "
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
-
"49": "
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
-
"58": "
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
|
@@ -227,12 +227,12 @@
|
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
-
"72": "
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
@@ -241,21 +241,21 @@
|
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NON_PARTICIPATING",
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
-
"91": "
|
| 252 |
-
"92": "
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
-
"98": "
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
@@ -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",
|
|
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "SUCCESS",
|
| 21 |
+
"113": "SUCCESS",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
|
|
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
+
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "SUCCESS",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
+
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
+
"144": "NOT_ALIVE",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
+
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
+
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "NOT_ALIVE",
|
| 66 |
+
"154": "SUCCESS",
|
| 67 |
+
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NOT_ALIVE",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
|
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
+
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NON_PARTICIPATING",
|
|
|
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
+
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
+
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
+
"214": "NOT_ALIVE",
|
| 134 |
+
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "NOT_ALIVE",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
|
|
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
+
"237": "NOT_ALIVE",
|
| 159 |
+
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
|
|
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
+
"247": "NOT_ALIVE",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NOT_ALIVE",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "SUCCESS",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
|
|
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
|
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
+
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
+
"49": "NOT_ALIVE",
|
| 205 |
+
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
+
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
+
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
+
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
|
|
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
+
"72": "NOT_ALIVE",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
|
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
+
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NON_PARTICIPATING",
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
+
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "SUCCESS",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "SUCCESS",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
+
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "NOT_ALIVE"
|
| 260 |
},
|
| 261 |
"architectures": [
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5991269,
|
| 268 |
+
5991300
|
| 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": 1,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5991043,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
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 sha256:5545d328272b966c92e805968d6079ebc42c365a456b00e9add3a61526cd77b6
|
| 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 sha256:8a9fdf253017af1f4614e837eee1462b3d6d1b9660b75a6eaf18d1a46fe220b2
|
| 3 |
size 4400216536
|