Run 11. Outer Step 209. Inner Step 9.
Browse files- config.json +40 -40
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -5,12 +5,12 @@
|
|
| 5 |
"1": "SUCCESS",
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
-
"101": "
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
-
"103": "
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
-
"106": "
|
| 14 |
"107": "SUCCESS",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "SUCCESS",
|
|
@@ -21,13 +21,13 @@
|
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "SUCCESS",
|
| 24 |
-
"116": "
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "SUCCESS",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
|
@@ -48,18 +48,18 @@
|
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
-
"140": "
|
| 52 |
"141": "SUCCESS",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "SUCCESS",
|
| 58 |
-
"147": "
|
| 59 |
-
"148": "
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "SUCCESS",
|
|
@@ -70,12 +70,12 @@
|
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "SUCCESS",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
-
"165": "
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NOT_ALIVE",
|
|
@@ -94,11 +94,11 @@
|
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
-
"182": "
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
-
"184": "
|
| 100 |
"185": "SUCCESS",
|
| 101 |
-
"186": "
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
|
@@ -106,7 +106,7 @@
|
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
-
"193": "
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
|
@@ -116,7 +116,7 @@
|
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
-
"201": "
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
|
@@ -126,28 +126,28 @@
|
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 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": "
|
| 141 |
-
"221": "
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "SUCCESS",
|
| 147 |
-
"227": "
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
-
"23": "
|
| 151 |
"230": "SUCCESS",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
|
@@ -155,28 +155,28 @@
|
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
-
"237": "
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NOT_ALIVE",
|
| 174 |
-
"251": "
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
-
"26": "
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
|
@@ -189,7 +189,7 @@
|
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "SUCCESS",
|
| 192 |
-
"38": "
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NOT_ALIVE",
|
|
@@ -200,11 +200,11 @@
|
|
| 200 |
"45": "SUCCESS",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
-
"48": "
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
-
"51": "
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
|
@@ -217,12 +217,12 @@
|
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
-
"68": "
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
|
@@ -247,16 +247,16 @@
|
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
-
"91": "
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
-
"97": "
|
| 258 |
"98": "SUCCESS",
|
| 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",
|
|
|
|
| 5 |
"1": "SUCCESS",
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
+
"101": "NOT_ALIVE",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
+
"103": "SUCCESS",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
+
"106": "FAIL",
|
| 14 |
"107": "SUCCESS",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "SUCCESS",
|
|
|
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "SUCCESS",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "SUCCESS",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
+
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
|
|
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
+
"140": "NOT_ALIVE",
|
| 52 |
"141": "SUCCESS",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "SUCCESS",
|
| 58 |
+
"147": "SUCCESS",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
+
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "SUCCESS",
|
|
|
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
+
"160": "SUCCESS",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "SUCCESS",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
+
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NOT_ALIVE",
|
|
|
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
+
"182": "SUCCESS",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
+
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "SUCCESS",
|
| 101 |
+
"186": "SUCCESS",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
|
|
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
+
"193": "SUCCESS",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
|
|
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
+
"201": "NOT_ALIVE",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
|
|
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "SUCCESS",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "NON_PARTICIPATING",
|
| 141 |
+
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "SUCCESS",
|
| 147 |
+
"227": "NOT_ALIVE",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
+
"23": "NOT_ALIVE",
|
| 151 |
"230": "SUCCESS",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
"232": "NOT_ALIVE",
|
|
|
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
+
"237": "SUCCESS",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "SUCCESS",
|
| 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": "SUCCESS",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
+
"26": "SUCCESS",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
|
|
|
| 189 |
"35": "NOT_ALIVE",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "SUCCESS",
|
| 192 |
+
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "NOT_ALIVE",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NOT_ALIVE",
|
|
|
|
| 200 |
"45": "SUCCESS",
|
| 201 |
"46": "NOT_ALIVE",
|
| 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",
|
|
|
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "SUCCESS",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
+
"63": "SUCCESS",
|
| 221 |
"64": "NOT_ALIVE",
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
+
"68": "NOT_ALIVE",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NOT_ALIVE",
|
|
|
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
+
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
+
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6662871,
|
| 268 |
+
6662891
|
| 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": 6662808,
|
| 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 88390091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96eb4b0222da47b0c5c04c149d0ce76ab8c70fc1a0645d22af509410a6412de3
|
| 3 |
size 88390091
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924928111
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f6132ad0c5b07c7145443072325d061ba47d4a2cc7554bad5d6aa17bac2e03aa
|
| 3 |
size 4924928111
|
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:e858c87b394b522b4f947ad2b32d04fdb7527964784bce34bffb5bccb9bc9d24
|
| 3 |
size 4400216536
|