Run 5. Outer Step 105. Inner Step 1.
Browse files- config.json +57 -57
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -17,11 +17,11 @@
|
|
| 17 |
"11": "NON_PARTICIPATING",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
-
"116": "
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
|
@@ -30,29 +30,29 @@
|
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
-
"124": "
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
-
"129": "
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "SUCCESS",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "SUCCESS",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
-
"14": "
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
-
"144": "
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
|
@@ -62,20 +62,20 @@
|
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
-
"157": "
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
|
@@ -83,7 +83,7 @@
|
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
-
"172": "
|
| 87 |
"173": "SUCCESS",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
|
@@ -95,7 +95,7 @@
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
|
@@ -105,11 +105,11 @@
|
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
-
"192": "
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
-
"194": "
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
-
"196": "
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
|
@@ -121,21 +121,21 @@
|
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
-
"213": "
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
-
"216": "
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NOT_ALIVE",
|
|
@@ -148,15 +148,15 @@
|
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
-
"235": "
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
|
@@ -177,56 +177,56 @@
|
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
-
"27": "
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
-
"29": "
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
-
"38": "
|
| 193 |
-
"39": "
|
| 194 |
-
"4": "
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
-
"41": "
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
-
"43": "
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
-
"48": "
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
-
"53": "
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
-
"6": "
|
| 217 |
-
"60": "
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
-
"65": "
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
-
"68": "
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
-
"71": "
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
|
@@ -247,15 +247,15 @@
|
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NOT_ALIVE",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
-
"96": "
|
| 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": "NON_PARTICIPATING",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NON_PARTICIPATING",
|
|
|
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
+
"124": "NOT_ALIVE",
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
+
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
+
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "SUCCESS",
|
| 43 |
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
+
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "SUCCESS",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "SUCCESS",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
+
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
+
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
|
|
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
+
"154": "SUCCESS",
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
+
"157": "NOT_ALIVE",
|
| 70 |
"158": "NON_PARTICIPATING",
|
| 71 |
+
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
|
|
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
+
"172": "SUCCESS",
|
| 87 |
"173": "SUCCESS",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
|
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NOT_ALIVE",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
|
|
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
+
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "NOT_ALIVE",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
+
"196": "NOT_ALIVE",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
|
|
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NOT_ALIVE",
|
| 135 |
+
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
+
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NOT_ALIVE",
|
|
|
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
+
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
+
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
|
|
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "SUCCESS",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
+
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "NOT_ALIVE",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
+
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
+
"41": "NOT_ALIVE",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
+
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
+
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "SUCCESS",
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
+
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "SUCCESS",
|
| 213 |
"57": "NOT_ALIVE",
|
| 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": "SUCCESS",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
+
"65": "NOT_ALIVE",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
+
"71": "NOT_ALIVE",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
|
|
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NOT_ALIVE",
|
| 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 |
+
5962433,
|
| 268 |
+
5962448
|
| 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": 5962363,
|
| 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:ddb07d23b7dff29d46187a54d7600477ee51a6fb6a9baeb98cda8a0c44f735fb
|
| 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:b9e7b9a34830f856bb37b1b82f007e968ab2de1e525d2edd045887b0171a559b
|
| 3 |
size 4400216536
|