Run 8. Outer Step 51. Inner Step 9.
Browse files- config.json +58 -58
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
@@ -8,10 +8,10 @@
|
|
8 |
"101": "NOT_ALIVE",
|
9 |
"102": "NON_PARTICIPATING",
|
10 |
"103": "NOT_ALIVE",
|
11 |
-
"104": "
|
12 |
"105": "NOT_ALIVE",
|
13 |
"106": "SUCCESS",
|
14 |
-
"107": "
|
15 |
"108": "NOT_ALIVE",
|
16 |
"109": "NON_PARTICIPATING",
|
17 |
"11": "NOT_ALIVE",
|
@@ -20,7 +20,7 @@
|
|
20 |
"112": "SUCCESS",
|
21 |
"113": "NOT_ALIVE",
|
22 |
"114": "NOT_ALIVE",
|
23 |
-
"115": "
|
24 |
"116": "NOT_ALIVE",
|
25 |
"117": "NOT_ALIVE",
|
26 |
"118": "NOT_ALIVE",
|
@@ -32,19 +32,19 @@
|
|
32 |
"123": "SUCCESS",
|
33 |
"124": "NOT_ALIVE",
|
34 |
"125": "NOT_ALIVE",
|
35 |
-
"126": "
|
36 |
-
"127": "
|
37 |
"128": "SUCCESS",
|
38 |
"129": "NON_PARTICIPATING",
|
39 |
"13": "NOT_ALIVE",
|
40 |
-
"130": "
|
41 |
"131": "NON_PARTICIPATING",
|
42 |
"132": "SUCCESS",
|
43 |
-
"133": "
|
44 |
"134": "NOT_ALIVE",
|
45 |
"135": "SUCCESS",
|
46 |
"136": "SUCCESS",
|
47 |
-
"137": "
|
48 |
"138": "NON_PARTICIPATING",
|
49 |
"139": "NOT_ALIVE",
|
50 |
"14": "NOT_ALIVE",
|
@@ -52,7 +52,7 @@
|
|
52 |
"141": "NOT_ALIVE",
|
53 |
"142": "NON_PARTICIPATING",
|
54 |
"143": "NON_PARTICIPATING",
|
55 |
-
"144": "
|
56 |
"145": "SUCCESS",
|
57 |
"146": "NON_PARTICIPATING",
|
58 |
"147": "NOT_ALIVE",
|
@@ -62,7 +62,7 @@
|
|
62 |
"150": "NOT_ALIVE",
|
63 |
"151": "NON_PARTICIPATING",
|
64 |
"152": "FAIL",
|
65 |
-
"153": "
|
66 |
"154": "NON_PARTICIPATING",
|
67 |
"155": "FAIL",
|
68 |
"156": "NOT_ALIVE",
|
@@ -73,7 +73,7 @@
|
|
73 |
"160": "NON_PARTICIPATING",
|
74 |
"161": "NON_PARTICIPATING",
|
75 |
"162": "NON_PARTICIPATING",
|
76 |
-
"163": "
|
77 |
"164": "FAIL",
|
78 |
"165": "SUCCESS",
|
79 |
"166": "NOT_ALIVE",
|
@@ -82,14 +82,14 @@
|
|
82 |
"169": "SUCCESS",
|
83 |
"17": "NOT_ALIVE",
|
84 |
"170": "SUCCESS",
|
85 |
-
"171": "
|
86 |
-
"172": "
|
87 |
-
"173": "
|
88 |
-
"174": "
|
89 |
-
"175": "
|
90 |
"176": "NOT_ALIVE",
|
91 |
"177": "NON_PARTICIPATING",
|
92 |
-
"178": "
|
93 |
"179": "SUCCESS",
|
94 |
"18": "NOT_ALIVE",
|
95 |
"180": "NON_PARTICIPATING",
|
@@ -98,12 +98,12 @@
|
|
98 |
"183": "NON_PARTICIPATING",
|
99 |
"184": "NOT_ALIVE",
|
100 |
"185": "NON_PARTICIPATING",
|
101 |
-
"186": "
|
102 |
"187": "NOT_ALIVE",
|
103 |
"188": "FAIL",
|
104 |
-
"189": "
|
105 |
"19": "NOT_ALIVE",
|
106 |
-
"190": "
|
107 |
"191": "FAIL",
|
108 |
"192": "FAIL",
|
109 |
"193": "NOT_ALIVE",
|
@@ -115,46 +115,46 @@
|
|
115 |
"199": "SUCCESS",
|
116 |
"2": "NOT_ALIVE",
|
117 |
"20": "NOT_ALIVE",
|
118 |
-
"200": "
|
119 |
"201": "NOT_ALIVE",
|
120 |
-
"202": "
|
121 |
"203": "NON_PARTICIPATING",
|
122 |
"204": "NON_PARTICIPATING",
|
123 |
"205": "NOT_ALIVE",
|
124 |
"206": "NOT_ALIVE",
|
125 |
"207": "NOT_ALIVE",
|
126 |
"208": "NOT_ALIVE",
|
127 |
-
"209": "
|
128 |
"21": "SUCCESS",
|
129 |
"210": "NOT_ALIVE",
|
130 |
"211": "NON_PARTICIPATING",
|
131 |
"212": "NOT_ALIVE",
|
132 |
"213": "NON_PARTICIPATING",
|
133 |
-
"214": "
|
134 |
-
"215": "
|
135 |
"216": "SUCCESS",
|
136 |
"217": "SUCCESS",
|
137 |
"218": "FAIL",
|
138 |
-
"219": "
|
139 |
"22": "NOT_ALIVE",
|
140 |
"220": "FAIL",
|
141 |
"221": "SUCCESS",
|
142 |
-
"222": "
|
143 |
"223": "NOT_ALIVE",
|
144 |
"224": "NON_PARTICIPATING",
|
145 |
"225": "NOT_ALIVE",
|
146 |
"226": "NOT_ALIVE",
|
147 |
-
"227": "
|
148 |
"228": "SUCCESS",
|
149 |
-
"229": "
|
150 |
"23": "NOT_ALIVE",
|
151 |
"230": "NON_PARTICIPATING",
|
152 |
-
"231": "
|
153 |
"232": "NOT_ALIVE",
|
154 |
"233": "NON_PARTICIPATING",
|
155 |
-
"234": "
|
156 |
"235": "NOT_ALIVE",
|
157 |
-
"236": "
|
158 |
"237": "NON_PARTICIPATING",
|
159 |
"238": "NOT_ALIVE",
|
160 |
"239": "NON_PARTICIPATING",
|
@@ -163,34 +163,34 @@
|
|
163 |
"241": "NON_PARTICIPATING",
|
164 |
"242": "NON_PARTICIPATING",
|
165 |
"243": "NON_PARTICIPATING",
|
166 |
-
"244": "
|
167 |
"245": "NON_PARTICIPATING",
|
168 |
"246": "NON_PARTICIPATING",
|
169 |
"247": "NON_PARTICIPATING",
|
170 |
-
"248": "
|
171 |
"249": "NOT_ALIVE",
|
172 |
"25": "SUCCESS",
|
173 |
"250": "NOT_ALIVE",
|
174 |
-
"251": "
|
175 |
-
"252": "
|
176 |
-
"253": "
|
177 |
"254": "NON_PARTICIPATING",
|
178 |
"255": "NON_PARTICIPATING",
|
179 |
"26": "FAIL",
|
180 |
"27": "SUCCESS",
|
181 |
"28": "NON_PARTICIPATING",
|
182 |
-
"29": "
|
183 |
"3": "FAIL",
|
184 |
-
"30": "
|
185 |
-
"31": "
|
186 |
"32": "NON_PARTICIPATING",
|
187 |
"33": "NOT_ALIVE",
|
188 |
"34": "NOT_ALIVE",
|
189 |
"35": "NON_PARTICIPATING",
|
190 |
"36": "FAIL",
|
191 |
"37": "NOT_ALIVE",
|
192 |
-
"38": "
|
193 |
-
"39": "
|
194 |
"4": "NOT_ALIVE",
|
195 |
"40": "NON_PARTICIPATING",
|
196 |
"41": "FAIL",
|
@@ -198,20 +198,20 @@
|
|
198 |
"43": "NON_PARTICIPATING",
|
199 |
"44": "FAIL",
|
200 |
"45": "NOT_ALIVE",
|
201 |
-
"46": "
|
202 |
"47": "NON_PARTICIPATING",
|
203 |
"48": "NON_PARTICIPATING",
|
204 |
"49": "NON_PARTICIPATING",
|
205 |
-
"5": "
|
206 |
"50": "NOT_ALIVE",
|
207 |
"51": "NON_PARTICIPATING",
|
208 |
-
"52": "
|
209 |
-
"53": "
|
210 |
-
"54": "
|
211 |
"55": "NON_PARTICIPATING",
|
212 |
"56": "NON_PARTICIPATING",
|
213 |
"57": "NOT_ALIVE",
|
214 |
-
"58": "
|
215 |
"59": "NOT_ALIVE",
|
216 |
"6": "NOT_ALIVE",
|
217 |
"60": "NOT_ALIVE",
|
@@ -234,12 +234,12 @@
|
|
234 |
"76": "NOT_ALIVE",
|
235 |
"77": "SUCCESS",
|
236 |
"78": "NON_PARTICIPATING",
|
237 |
-
"79": "
|
238 |
"8": "NOT_ALIVE",
|
239 |
"80": "NON_PARTICIPATING",
|
240 |
"81": "NON_PARTICIPATING",
|
241 |
-
"82": "
|
242 |
-
"83": "
|
243 |
"84": "FAIL",
|
244 |
"85": "NOT_ALIVE",
|
245 |
"86": "NOT_ALIVE",
|
@@ -253,9 +253,9 @@
|
|
253 |
"93": "NOT_ALIVE",
|
254 |
"94": "NOT_ALIVE",
|
255 |
"95": "NOT_ALIVE",
|
256 |
-
"96": "
|
257 |
-
"97": "
|
258 |
-
"98": "
|
259 |
"99": "NON_PARTICIPATING"
|
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",
|
|
|
8 |
"101": "NOT_ALIVE",
|
9 |
"102": "NON_PARTICIPATING",
|
10 |
"103": "NOT_ALIVE",
|
11 |
+
"104": "NON_PARTICIPATING",
|
12 |
"105": "NOT_ALIVE",
|
13 |
"106": "SUCCESS",
|
14 |
+
"107": "NON_PARTICIPATING",
|
15 |
"108": "NOT_ALIVE",
|
16 |
"109": "NON_PARTICIPATING",
|
17 |
"11": "NOT_ALIVE",
|
|
|
20 |
"112": "SUCCESS",
|
21 |
"113": "NOT_ALIVE",
|
22 |
"114": "NOT_ALIVE",
|
23 |
+
"115": "NOT_ALIVE",
|
24 |
"116": "NOT_ALIVE",
|
25 |
"117": "NOT_ALIVE",
|
26 |
"118": "NOT_ALIVE",
|
|
|
32 |
"123": "SUCCESS",
|
33 |
"124": "NOT_ALIVE",
|
34 |
"125": "NOT_ALIVE",
|
35 |
+
"126": "SUCCESS",
|
36 |
+
"127": "SUCCESS",
|
37 |
"128": "SUCCESS",
|
38 |
"129": "NON_PARTICIPATING",
|
39 |
"13": "NOT_ALIVE",
|
40 |
+
"130": "NOT_ALIVE",
|
41 |
"131": "NON_PARTICIPATING",
|
42 |
"132": "SUCCESS",
|
43 |
+
"133": "SUCCESS",
|
44 |
"134": "NOT_ALIVE",
|
45 |
"135": "SUCCESS",
|
46 |
"136": "SUCCESS",
|
47 |
+
"137": "NON_PARTICIPATING",
|
48 |
"138": "NON_PARTICIPATING",
|
49 |
"139": "NOT_ALIVE",
|
50 |
"14": "NOT_ALIVE",
|
|
|
52 |
"141": "NOT_ALIVE",
|
53 |
"142": "NON_PARTICIPATING",
|
54 |
"143": "NON_PARTICIPATING",
|
55 |
+
"144": "SUCCESS",
|
56 |
"145": "SUCCESS",
|
57 |
"146": "NON_PARTICIPATING",
|
58 |
"147": "NOT_ALIVE",
|
|
|
62 |
"150": "NOT_ALIVE",
|
63 |
"151": "NON_PARTICIPATING",
|
64 |
"152": "FAIL",
|
65 |
+
"153": "SUCCESS",
|
66 |
"154": "NON_PARTICIPATING",
|
67 |
"155": "FAIL",
|
68 |
"156": "NOT_ALIVE",
|
|
|
73 |
"160": "NON_PARTICIPATING",
|
74 |
"161": "NON_PARTICIPATING",
|
75 |
"162": "NON_PARTICIPATING",
|
76 |
+
"163": "FAIL",
|
77 |
"164": "FAIL",
|
78 |
"165": "SUCCESS",
|
79 |
"166": "NOT_ALIVE",
|
|
|
82 |
"169": "SUCCESS",
|
83 |
"17": "NOT_ALIVE",
|
84 |
"170": "SUCCESS",
|
85 |
+
"171": "SUCCESS",
|
86 |
+
"172": "NON_PARTICIPATING",
|
87 |
+
"173": "NON_PARTICIPATING",
|
88 |
+
"174": "NON_PARTICIPATING",
|
89 |
+
"175": "NON_PARTICIPATING",
|
90 |
"176": "NOT_ALIVE",
|
91 |
"177": "NON_PARTICIPATING",
|
92 |
+
"178": "NON_PARTICIPATING",
|
93 |
"179": "SUCCESS",
|
94 |
"18": "NOT_ALIVE",
|
95 |
"180": "NON_PARTICIPATING",
|
|
|
98 |
"183": "NON_PARTICIPATING",
|
99 |
"184": "NOT_ALIVE",
|
100 |
"185": "NON_PARTICIPATING",
|
101 |
+
"186": "FAIL",
|
102 |
"187": "NOT_ALIVE",
|
103 |
"188": "FAIL",
|
104 |
+
"189": "SUCCESS",
|
105 |
"19": "NOT_ALIVE",
|
106 |
+
"190": "NON_PARTICIPATING",
|
107 |
"191": "FAIL",
|
108 |
"192": "FAIL",
|
109 |
"193": "NOT_ALIVE",
|
|
|
115 |
"199": "SUCCESS",
|
116 |
"2": "NOT_ALIVE",
|
117 |
"20": "NOT_ALIVE",
|
118 |
+
"200": "NON_PARTICIPATING",
|
119 |
"201": "NOT_ALIVE",
|
120 |
+
"202": "FAIL",
|
121 |
"203": "NON_PARTICIPATING",
|
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": "SUCCESS",
|
129 |
"210": "NOT_ALIVE",
|
130 |
"211": "NON_PARTICIPATING",
|
131 |
"212": "NOT_ALIVE",
|
132 |
"213": "NON_PARTICIPATING",
|
133 |
+
"214": "NON_PARTICIPATING",
|
134 |
+
"215": "FAIL",
|
135 |
"216": "SUCCESS",
|
136 |
"217": "SUCCESS",
|
137 |
"218": "FAIL",
|
138 |
+
"219": "NON_PARTICIPATING",
|
139 |
"22": "NOT_ALIVE",
|
140 |
"220": "FAIL",
|
141 |
"221": "SUCCESS",
|
142 |
+
"222": "SUCCESS",
|
143 |
"223": "NOT_ALIVE",
|
144 |
"224": "NON_PARTICIPATING",
|
145 |
"225": "NOT_ALIVE",
|
146 |
"226": "NOT_ALIVE",
|
147 |
+
"227": "NON_PARTICIPATING",
|
148 |
"228": "SUCCESS",
|
149 |
+
"229": "NOT_ALIVE",
|
150 |
"23": "NOT_ALIVE",
|
151 |
"230": "NON_PARTICIPATING",
|
152 |
+
"231": "NOT_ALIVE",
|
153 |
"232": "NOT_ALIVE",
|
154 |
"233": "NON_PARTICIPATING",
|
155 |
+
"234": "FAIL",
|
156 |
"235": "NOT_ALIVE",
|
157 |
+
"236": "SUCCESS",
|
158 |
"237": "NON_PARTICIPATING",
|
159 |
"238": "NOT_ALIVE",
|
160 |
"239": "NON_PARTICIPATING",
|
|
|
163 |
"241": "NON_PARTICIPATING",
|
164 |
"242": "NON_PARTICIPATING",
|
165 |
"243": "NON_PARTICIPATING",
|
166 |
+
"244": "NOT_ALIVE",
|
167 |
"245": "NON_PARTICIPATING",
|
168 |
"246": "NON_PARTICIPATING",
|
169 |
"247": "NON_PARTICIPATING",
|
170 |
+
"248": "NOT_ALIVE",
|
171 |
"249": "NOT_ALIVE",
|
172 |
"25": "SUCCESS",
|
173 |
"250": "NOT_ALIVE",
|
174 |
+
"251": "NON_PARTICIPATING",
|
175 |
+
"252": "NOT_ALIVE",
|
176 |
+
"253": "NON_PARTICIPATING",
|
177 |
"254": "NON_PARTICIPATING",
|
178 |
"255": "NON_PARTICIPATING",
|
179 |
"26": "FAIL",
|
180 |
"27": "SUCCESS",
|
181 |
"28": "NON_PARTICIPATING",
|
182 |
+
"29": "NOT_ALIVE",
|
183 |
"3": "FAIL",
|
184 |
+
"30": "NON_PARTICIPATING",
|
185 |
+
"31": "SUCCESS",
|
186 |
"32": "NON_PARTICIPATING",
|
187 |
"33": "NOT_ALIVE",
|
188 |
"34": "NOT_ALIVE",
|
189 |
"35": "NON_PARTICIPATING",
|
190 |
"36": "FAIL",
|
191 |
"37": "NOT_ALIVE",
|
192 |
+
"38": "NOT_ALIVE",
|
193 |
+
"39": "NOT_ALIVE",
|
194 |
"4": "NOT_ALIVE",
|
195 |
"40": "NON_PARTICIPATING",
|
196 |
"41": "FAIL",
|
|
|
198 |
"43": "NON_PARTICIPATING",
|
199 |
"44": "FAIL",
|
200 |
"45": "NOT_ALIVE",
|
201 |
+
"46": "SUCCESS",
|
202 |
"47": "NON_PARTICIPATING",
|
203 |
"48": "NON_PARTICIPATING",
|
204 |
"49": "NON_PARTICIPATING",
|
205 |
+
"5": "FAIL",
|
206 |
"50": "NOT_ALIVE",
|
207 |
"51": "NON_PARTICIPATING",
|
208 |
+
"52": "NON_PARTICIPATING",
|
209 |
+
"53": "FAIL",
|
210 |
+
"54": "NOT_ALIVE",
|
211 |
"55": "NON_PARTICIPATING",
|
212 |
"56": "NON_PARTICIPATING",
|
213 |
"57": "NOT_ALIVE",
|
214 |
+
"58": "NOT_ALIVE",
|
215 |
"59": "NOT_ALIVE",
|
216 |
"6": "NOT_ALIVE",
|
217 |
"60": "NOT_ALIVE",
|
|
|
234 |
"76": "NOT_ALIVE",
|
235 |
"77": "SUCCESS",
|
236 |
"78": "NON_PARTICIPATING",
|
237 |
+
"79": "SUCCESS",
|
238 |
"8": "NOT_ALIVE",
|
239 |
"80": "NON_PARTICIPATING",
|
240 |
"81": "NON_PARTICIPATING",
|
241 |
+
"82": "SUCCESS",
|
242 |
+
"83": "NOT_ALIVE",
|
243 |
"84": "FAIL",
|
244 |
"85": "NOT_ALIVE",
|
245 |
"86": "NOT_ALIVE",
|
|
|
253 |
"93": "NOT_ALIVE",
|
254 |
"94": "NOT_ALIVE",
|
255 |
"95": "NOT_ALIVE",
|
256 |
+
"96": "NOT_ALIVE",
|
257 |
+
"97": "NOT_ALIVE",
|
258 |
+
"98": "FAIL",
|
259 |
"99": "NON_PARTICIPATING"
|
260 |
},
|
261 |
"architectures": [
|
|
|
264 |
"attention_bias": false,
|
265 |
"attention_dropout": 0.0,
|
266 |
"block_list": [
|
267 |
+
6374868,
|
268 |
+
6374894
|
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": 9,
|
276 |
"intermediate_size": 5632,
|
277 |
+
"last_allreduce_block": 6374792,
|
278 |
"max_position_embeddings": 2048,
|
279 |
"mlp_bias": false,
|
280 |
"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:d12a7b2d024992430ffa281e8a94a96815da693e45d1338694b554e2dd0f095e
|
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 sha256:274a560892c74f3d71415e61ac6ec52f20b99a68552d6518e051dd129d646728
|
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:dc5b328691b8ae93f7c5d04fdc6b13702dfe2e9cbdab0c4e64d792ddfe454a65
|
3 |
size 4400216536
|