Epoch 1105. Batch Size 55784. Peers 99.
Browse files- config.json +112 -112
- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- optimizer.pt +1 -1
config.json
CHANGED
@@ -3,64 +3,64 @@
|
|
3 |
"activation_function": "gelu_new",
|
4 |
"all_reduce_scores": {
|
5 |
"0": "NON_PARTICIPATING",
|
6 |
-
"1": "
|
7 |
"10": "NON_PARTICIPATING",
|
8 |
"100": "NON_PARTICIPATING",
|
9 |
"101": "NON_PARTICIPATING",
|
10 |
"102": "NON_PARTICIPATING",
|
11 |
-
"103": "
|
12 |
-
"104": "
|
13 |
-
"105": "
|
14 |
-
"106": "
|
15 |
-
"107": "
|
16 |
-
"108": "
|
17 |
-
"109": "
|
18 |
"11": "NON_PARTICIPATING",
|
19 |
"110": "NON_PARTICIPATING",
|
20 |
-
"111": "
|
21 |
-
"112": "
|
22 |
"113": "NON_PARTICIPATING",
|
23 |
-
"114": "
|
24 |
"115": "NON_PARTICIPATING",
|
25 |
"116": "NON_PARTICIPATING",
|
26 |
-
"117": "
|
27 |
"118": "NON_PARTICIPATING",
|
28 |
"119": "NON_PARTICIPATING",
|
29 |
"12": "NON_PARTICIPATING",
|
30 |
-
"120": "
|
31 |
-
"121": "
|
32 |
"122": "SUCCESS",
|
33 |
-
"123": "
|
34 |
"124": "NON_PARTICIPATING",
|
35 |
"125": "NON_PARTICIPATING",
|
36 |
"126": "NON_PARTICIPATING",
|
37 |
"127": "NON_PARTICIPATING",
|
38 |
-
"128": "
|
39 |
-
"129": "
|
40 |
-
"13": "
|
41 |
-
"130": "
|
42 |
"131": "NON_PARTICIPATING",
|
43 |
"132": "SUCCESS",
|
44 |
-
"133": "
|
45 |
"134": "NON_PARTICIPATING",
|
46 |
"135": "NON_PARTICIPATING",
|
47 |
-
"136": "
|
48 |
-
"137": "
|
49 |
"138": "NON_PARTICIPATING",
|
50 |
-
"139": "
|
51 |
"14": "NON_PARTICIPATING",
|
52 |
"140": "SUCCESS",
|
53 |
-
"141": "
|
54 |
-
"142": "
|
55 |
"143": "NON_PARTICIPATING",
|
56 |
"144": "NON_PARTICIPATING",
|
57 |
-
"145": "
|
58 |
"146": "NON_PARTICIPATING",
|
59 |
"147": "NON_PARTICIPATING",
|
60 |
-
"148": "
|
61 |
"149": "NON_PARTICIPATING",
|
62 |
"15": "NON_PARTICIPATING",
|
63 |
-
"150": "
|
64 |
"151": "NON_PARTICIPATING",
|
65 |
"152": "NON_PARTICIPATING",
|
66 |
"153": "SUCCESS",
|
@@ -68,102 +68,102 @@
|
|
68 |
"155": "SUCCESS",
|
69 |
"156": "SUCCESS",
|
70 |
"157": "NON_PARTICIPATING",
|
71 |
-
"158": "
|
72 |
-
"159": "
|
73 |
-
"16": "
|
74 |
"160": "SUCCESS",
|
75 |
"161": "SUCCESS",
|
76 |
-
"162": "
|
77 |
"163": "NON_PARTICIPATING",
|
78 |
-
"164": "
|
79 |
"165": "NON_PARTICIPATING",
|
80 |
-
"166": "
|
81 |
-
"167": "
|
82 |
-
"168": "
|
83 |
-
"169": "
|
84 |
-
"17": "
|
85 |
-
"170": "
|
86 |
"171": "SUCCESS",
|
87 |
"172": "SUCCESS",
|
88 |
"173": "SUCCESS",
|
89 |
"174": "NON_PARTICIPATING",
|
90 |
"175": "SUCCESS",
|
91 |
-
"176": "
|
92 |
-
"177": "
|
93 |
"178": "NON_PARTICIPATING",
|
94 |
-
"179": "
|
95 |
"18": "SUCCESS",
|
96 |
"180": "SUCCESS",
|
97 |
-
"181": "
|
98 |
"182": "NON_PARTICIPATING",
|
99 |
"183": "NON_PARTICIPATING",
|
100 |
-
"184": "
|
101 |
"185": "NON_PARTICIPATING",
|
102 |
-
"186": "
|
103 |
"187": "NON_PARTICIPATING",
|
104 |
"188": "NON_PARTICIPATING",
|
105 |
"189": "NON_PARTICIPATING",
|
106 |
-
"19": "
|
107 |
"190": "NON_PARTICIPATING",
|
108 |
"191": "SUCCESS",
|
109 |
"192": "NON_PARTICIPATING",
|
110 |
"193": "NON_PARTICIPATING",
|
111 |
"194": "NON_PARTICIPATING",
|
112 |
"195": "SUCCESS",
|
113 |
-
"196": "
|
114 |
-
"197": "
|
115 |
"198": "SUCCESS",
|
116 |
-
"199": "
|
117 |
-
"2": "
|
118 |
"20": "NON_PARTICIPATING",
|
119 |
-
"200": "
|
120 |
"201": "NON_PARTICIPATING",
|
121 |
-
"202": "
|
122 |
"203": "SUCCESS",
|
123 |
"204": "SUCCESS",
|
124 |
"205": "SUCCESS",
|
125 |
"206": "SUCCESS",
|
126 |
"207": "NON_PARTICIPATING",
|
127 |
-
"208": "
|
128 |
-
"209": "
|
129 |
-
"21": "
|
130 |
-
"210": "
|
131 |
"211": "SUCCESS",
|
132 |
"212": "SUCCESS",
|
133 |
"213": "SUCCESS",
|
134 |
-
"214": "
|
135 |
-
"215": "
|
136 |
"216": "NON_PARTICIPATING",
|
137 |
"217": "SUCCESS",
|
138 |
-
"218": "
|
139 |
"219": "NON_PARTICIPATING",
|
140 |
-
"22": "
|
141 |
"220": "NON_PARTICIPATING",
|
142 |
"221": "NON_PARTICIPATING",
|
143 |
"222": "NON_PARTICIPATING",
|
144 |
"223": "NON_PARTICIPATING",
|
145 |
"224": "NON_PARTICIPATING",
|
146 |
-
"225": "
|
147 |
"226": "SUCCESS",
|
148 |
"227": "NON_PARTICIPATING",
|
149 |
"228": "NON_PARTICIPATING",
|
150 |
"229": "NON_PARTICIPATING",
|
151 |
"23": "SUCCESS",
|
152 |
-
"230": "
|
153 |
"231": "NON_PARTICIPATING",
|
154 |
-
"232": "
|
155 |
"233": "NON_PARTICIPATING",
|
156 |
-
"234": "
|
157 |
-
"235": "
|
158 |
-
"236": "
|
159 |
"237": "NON_PARTICIPATING",
|
160 |
"238": "NON_PARTICIPATING",
|
161 |
"239": "NON_PARTICIPATING",
|
162 |
"24": "NON_PARTICIPATING",
|
163 |
-
"240": "
|
164 |
"241": "NON_PARTICIPATING",
|
165 |
-
"242": "
|
166 |
-
"243": "
|
167 |
"244": "SUCCESS",
|
168 |
"245": "NON_PARTICIPATING",
|
169 |
"246": "SUCCESS",
|
@@ -171,85 +171,85 @@
|
|
171 |
"248": "NON_PARTICIPATING",
|
172 |
"249": "NON_PARTICIPATING",
|
173 |
"25": "NON_PARTICIPATING",
|
174 |
-
"250": "
|
175 |
"251": "SUCCESS",
|
176 |
-
"252": "
|
177 |
"253": "NON_PARTICIPATING",
|
178 |
-
"254": "
|
179 |
"255": "NON_PARTICIPATING",
|
180 |
-
"26": "
|
181 |
"27": "NON_PARTICIPATING",
|
182 |
-
"28": "
|
183 |
"29": "SUCCESS",
|
184 |
-
"3": "
|
185 |
-
"30": "
|
186 |
"31": "SUCCESS",
|
187 |
"32": "SUCCESS",
|
188 |
"33": "NON_PARTICIPATING",
|
189 |
"34": "SUCCESS",
|
190 |
-
"35": "
|
191 |
-
"36": "
|
192 |
-
"37": "
|
193 |
"38": "NON_PARTICIPATING",
|
194 |
-
"39": "
|
195 |
-
"4": "
|
196 |
"40": "NON_PARTICIPATING",
|
197 |
"41": "NON_PARTICIPATING",
|
198 |
-
"42": "
|
199 |
"43": "NON_PARTICIPATING",
|
200 |
-
"44": "
|
201 |
-
"45": "
|
202 |
-
"46": "
|
203 |
"47": "NON_PARTICIPATING",
|
204 |
-
"48": "
|
205 |
-
"49": "
|
206 |
-
"5": "
|
207 |
-
"50": "
|
208 |
"51": "NON_PARTICIPATING",
|
209 |
-
"52": "
|
210 |
-
"53": "
|
211 |
-
"54": "
|
212 |
"55": "SUCCESS",
|
213 |
"56": "SUCCESS",
|
214 |
"57": "NON_PARTICIPATING",
|
215 |
-
"58": "
|
216 |
"59": "NON_PARTICIPATING",
|
217 |
-
"6": "
|
218 |
"60": "SUCCESS",
|
219 |
-
"61": "
|
220 |
"62": "NON_PARTICIPATING",
|
221 |
"63": "NON_PARTICIPATING",
|
222 |
"64": "NON_PARTICIPATING",
|
223 |
"65": "SUCCESS",
|
224 |
-
"66": "
|
225 |
"67": "NON_PARTICIPATING",
|
226 |
-
"68": "
|
227 |
"69": "NON_PARTICIPATING",
|
228 |
-
"7": "
|
229 |
-
"70": "
|
230 |
-
"71": "
|
231 |
-
"72": "
|
232 |
"73": "NON_PARTICIPATING",
|
233 |
"74": "NON_PARTICIPATING",
|
234 |
-
"75": "
|
235 |
"76": "NON_PARTICIPATING",
|
236 |
"77": "NON_PARTICIPATING",
|
237 |
-
"78": "
|
238 |
-
"79": "
|
239 |
"8": "NON_PARTICIPATING",
|
240 |
-
"80": "
|
241 |
"81": "NON_PARTICIPATING",
|
242 |
-
"82": "
|
243 |
"83": "SUCCESS",
|
244 |
"84": "NON_PARTICIPATING",
|
245 |
-
"85": "
|
246 |
-
"86": "
|
247 |
"87": "NON_PARTICIPATING",
|
248 |
-
"88": "
|
249 |
"89": "SUCCESS",
|
250 |
"9": "NON_PARTICIPATING",
|
251 |
-
"90": "
|
252 |
-
"91": "
|
253 |
"92": "NON_PARTICIPATING",
|
254 |
"93": "NON_PARTICIPATING",
|
255 |
"94": "SUCCESS",
|
@@ -257,7 +257,7 @@
|
|
257 |
"96": "NON_PARTICIPATING",
|
258 |
"97": "NON_PARTICIPATING",
|
259 |
"98": "NON_PARTICIPATING",
|
260 |
-
"99": "
|
261 |
},
|
262 |
"architectures": [
|
263 |
"GPTOptim"
|
|
|
3 |
"activation_function": "gelu_new",
|
4 |
"all_reduce_scores": {
|
5 |
"0": "NON_PARTICIPATING",
|
6 |
+
"1": "SUCCESS",
|
7 |
"10": "NON_PARTICIPATING",
|
8 |
"100": "NON_PARTICIPATING",
|
9 |
"101": "NON_PARTICIPATING",
|
10 |
"102": "NON_PARTICIPATING",
|
11 |
+
"103": "SUCCESS",
|
12 |
+
"104": "NON_PARTICIPATING",
|
13 |
+
"105": "SUCCESS",
|
14 |
+
"106": "SUCCESS",
|
15 |
+
"107": "SUCCESS",
|
16 |
+
"108": "NON_PARTICIPATING",
|
17 |
+
"109": "SUCCESS",
|
18 |
"11": "NON_PARTICIPATING",
|
19 |
"110": "NON_PARTICIPATING",
|
20 |
+
"111": "SUCCESS",
|
21 |
+
"112": "SUCCESS",
|
22 |
"113": "NON_PARTICIPATING",
|
23 |
+
"114": "NON_PARTICIPATING",
|
24 |
"115": "NON_PARTICIPATING",
|
25 |
"116": "NON_PARTICIPATING",
|
26 |
+
"117": "NON_PARTICIPATING",
|
27 |
"118": "NON_PARTICIPATING",
|
28 |
"119": "NON_PARTICIPATING",
|
29 |
"12": "NON_PARTICIPATING",
|
30 |
+
"120": "NON_PARTICIPATING",
|
31 |
+
"121": "SUCCESS",
|
32 |
"122": "SUCCESS",
|
33 |
+
"123": "NON_PARTICIPATING",
|
34 |
"124": "NON_PARTICIPATING",
|
35 |
"125": "NON_PARTICIPATING",
|
36 |
"126": "NON_PARTICIPATING",
|
37 |
"127": "NON_PARTICIPATING",
|
38 |
+
"128": "SUCCESS",
|
39 |
+
"129": "NON_PARTICIPATING",
|
40 |
+
"13": "NON_PARTICIPATING",
|
41 |
+
"130": "NON_PARTICIPATING",
|
42 |
"131": "NON_PARTICIPATING",
|
43 |
"132": "SUCCESS",
|
44 |
+
"133": "NON_PARTICIPATING",
|
45 |
"134": "NON_PARTICIPATING",
|
46 |
"135": "NON_PARTICIPATING",
|
47 |
+
"136": "SUCCESS",
|
48 |
+
"137": "SUCCESS",
|
49 |
"138": "NON_PARTICIPATING",
|
50 |
+
"139": "NON_PARTICIPATING",
|
51 |
"14": "NON_PARTICIPATING",
|
52 |
"140": "SUCCESS",
|
53 |
+
"141": "SUCCESS",
|
54 |
+
"142": "SUCCESS",
|
55 |
"143": "NON_PARTICIPATING",
|
56 |
"144": "NON_PARTICIPATING",
|
57 |
+
"145": "SUCCESS",
|
58 |
"146": "NON_PARTICIPATING",
|
59 |
"147": "NON_PARTICIPATING",
|
60 |
+
"148": "NON_PARTICIPATING",
|
61 |
"149": "NON_PARTICIPATING",
|
62 |
"15": "NON_PARTICIPATING",
|
63 |
+
"150": "SUCCESS",
|
64 |
"151": "NON_PARTICIPATING",
|
65 |
"152": "NON_PARTICIPATING",
|
66 |
"153": "SUCCESS",
|
|
|
68 |
"155": "SUCCESS",
|
69 |
"156": "SUCCESS",
|
70 |
"157": "NON_PARTICIPATING",
|
71 |
+
"158": "NON_PARTICIPATING",
|
72 |
+
"159": "NON_PARTICIPATING",
|
73 |
+
"16": "NON_PARTICIPATING",
|
74 |
"160": "SUCCESS",
|
75 |
"161": "SUCCESS",
|
76 |
+
"162": "NON_PARTICIPATING",
|
77 |
"163": "NON_PARTICIPATING",
|
78 |
+
"164": "SUCCESS",
|
79 |
"165": "NON_PARTICIPATING",
|
80 |
+
"166": "SUCCESS",
|
81 |
+
"167": "SUCCESS",
|
82 |
+
"168": "NON_PARTICIPATING",
|
83 |
+
"169": "SUCCESS",
|
84 |
+
"17": "SUCCESS",
|
85 |
+
"170": "SUCCESS",
|
86 |
"171": "SUCCESS",
|
87 |
"172": "SUCCESS",
|
88 |
"173": "SUCCESS",
|
89 |
"174": "NON_PARTICIPATING",
|
90 |
"175": "SUCCESS",
|
91 |
+
"176": "NON_PARTICIPATING",
|
92 |
+
"177": "SUCCESS",
|
93 |
"178": "NON_PARTICIPATING",
|
94 |
+
"179": "SUCCESS",
|
95 |
"18": "SUCCESS",
|
96 |
"180": "SUCCESS",
|
97 |
+
"181": "NON_PARTICIPATING",
|
98 |
"182": "NON_PARTICIPATING",
|
99 |
"183": "NON_PARTICIPATING",
|
100 |
+
"184": "SUCCESS",
|
101 |
"185": "NON_PARTICIPATING",
|
102 |
+
"186": "SUCCESS",
|
103 |
"187": "NON_PARTICIPATING",
|
104 |
"188": "NON_PARTICIPATING",
|
105 |
"189": "NON_PARTICIPATING",
|
106 |
+
"19": "SUCCESS",
|
107 |
"190": "NON_PARTICIPATING",
|
108 |
"191": "SUCCESS",
|
109 |
"192": "NON_PARTICIPATING",
|
110 |
"193": "NON_PARTICIPATING",
|
111 |
"194": "NON_PARTICIPATING",
|
112 |
"195": "SUCCESS",
|
113 |
+
"196": "NON_PARTICIPATING",
|
114 |
+
"197": "SUCCESS",
|
115 |
"198": "SUCCESS",
|
116 |
+
"199": "SUCCESS",
|
117 |
+
"2": "NON_PARTICIPATING",
|
118 |
"20": "NON_PARTICIPATING",
|
119 |
+
"200": "SUCCESS",
|
120 |
"201": "NON_PARTICIPATING",
|
121 |
+
"202": "NON_PARTICIPATING",
|
122 |
"203": "SUCCESS",
|
123 |
"204": "SUCCESS",
|
124 |
"205": "SUCCESS",
|
125 |
"206": "SUCCESS",
|
126 |
"207": "NON_PARTICIPATING",
|
127 |
+
"208": "SUCCESS",
|
128 |
+
"209": "SUCCESS",
|
129 |
+
"21": "NON_PARTICIPATING",
|
130 |
+
"210": "NON_PARTICIPATING",
|
131 |
"211": "SUCCESS",
|
132 |
"212": "SUCCESS",
|
133 |
"213": "SUCCESS",
|
134 |
+
"214": "NON_PARTICIPATING",
|
135 |
+
"215": "NON_PARTICIPATING",
|
136 |
"216": "NON_PARTICIPATING",
|
137 |
"217": "SUCCESS",
|
138 |
+
"218": "SUCCESS",
|
139 |
"219": "NON_PARTICIPATING",
|
140 |
+
"22": "SUCCESS",
|
141 |
"220": "NON_PARTICIPATING",
|
142 |
"221": "NON_PARTICIPATING",
|
143 |
"222": "NON_PARTICIPATING",
|
144 |
"223": "NON_PARTICIPATING",
|
145 |
"224": "NON_PARTICIPATING",
|
146 |
+
"225": "NON_PARTICIPATING",
|
147 |
"226": "SUCCESS",
|
148 |
"227": "NON_PARTICIPATING",
|
149 |
"228": "NON_PARTICIPATING",
|
150 |
"229": "NON_PARTICIPATING",
|
151 |
"23": "SUCCESS",
|
152 |
+
"230": "SUCCESS",
|
153 |
"231": "NON_PARTICIPATING",
|
154 |
+
"232": "SUCCESS",
|
155 |
"233": "NON_PARTICIPATING",
|
156 |
+
"234": "NON_PARTICIPATING",
|
157 |
+
"235": "NON_PARTICIPATING",
|
158 |
+
"236": "SUCCESS",
|
159 |
"237": "NON_PARTICIPATING",
|
160 |
"238": "NON_PARTICIPATING",
|
161 |
"239": "NON_PARTICIPATING",
|
162 |
"24": "NON_PARTICIPATING",
|
163 |
+
"240": "SUCCESS",
|
164 |
"241": "NON_PARTICIPATING",
|
165 |
+
"242": "NON_PARTICIPATING",
|
166 |
+
"243": "SUCCESS",
|
167 |
"244": "SUCCESS",
|
168 |
"245": "NON_PARTICIPATING",
|
169 |
"246": "SUCCESS",
|
|
|
171 |
"248": "NON_PARTICIPATING",
|
172 |
"249": "NON_PARTICIPATING",
|
173 |
"25": "NON_PARTICIPATING",
|
174 |
+
"250": "NON_PARTICIPATING",
|
175 |
"251": "SUCCESS",
|
176 |
+
"252": "NON_PARTICIPATING",
|
177 |
"253": "NON_PARTICIPATING",
|
178 |
+
"254": "NON_PARTICIPATING",
|
179 |
"255": "NON_PARTICIPATING",
|
180 |
+
"26": "SUCCESS",
|
181 |
"27": "NON_PARTICIPATING",
|
182 |
+
"28": "SUCCESS",
|
183 |
"29": "SUCCESS",
|
184 |
+
"3": "NON_PARTICIPATING",
|
185 |
+
"30": "SUCCESS",
|
186 |
"31": "SUCCESS",
|
187 |
"32": "SUCCESS",
|
188 |
"33": "NON_PARTICIPATING",
|
189 |
"34": "SUCCESS",
|
190 |
+
"35": "NON_PARTICIPATING",
|
191 |
+
"36": "SUCCESS",
|
192 |
+
"37": "NON_PARTICIPATING",
|
193 |
"38": "NON_PARTICIPATING",
|
194 |
+
"39": "NON_PARTICIPATING",
|
195 |
+
"4": "NON_PARTICIPATING",
|
196 |
"40": "NON_PARTICIPATING",
|
197 |
"41": "NON_PARTICIPATING",
|
198 |
+
"42": "SUCCESS",
|
199 |
"43": "NON_PARTICIPATING",
|
200 |
+
"44": "SUCCESS",
|
201 |
+
"45": "NON_PARTICIPATING",
|
202 |
+
"46": "NON_PARTICIPATING",
|
203 |
"47": "NON_PARTICIPATING",
|
204 |
+
"48": "NON_PARTICIPATING",
|
205 |
+
"49": "SUCCESS",
|
206 |
+
"5": "SUCCESS",
|
207 |
+
"50": "SUCCESS",
|
208 |
"51": "NON_PARTICIPATING",
|
209 |
+
"52": "SUCCESS",
|
210 |
+
"53": "NON_PARTICIPATING",
|
211 |
+
"54": "NON_PARTICIPATING",
|
212 |
"55": "SUCCESS",
|
213 |
"56": "SUCCESS",
|
214 |
"57": "NON_PARTICIPATING",
|
215 |
+
"58": "NON_PARTICIPATING",
|
216 |
"59": "NON_PARTICIPATING",
|
217 |
+
"6": "NON_PARTICIPATING",
|
218 |
"60": "SUCCESS",
|
219 |
+
"61": "NON_PARTICIPATING",
|
220 |
"62": "NON_PARTICIPATING",
|
221 |
"63": "NON_PARTICIPATING",
|
222 |
"64": "NON_PARTICIPATING",
|
223 |
"65": "SUCCESS",
|
224 |
+
"66": "NON_PARTICIPATING",
|
225 |
"67": "NON_PARTICIPATING",
|
226 |
+
"68": "NON_PARTICIPATING",
|
227 |
"69": "NON_PARTICIPATING",
|
228 |
+
"7": "NON_PARTICIPATING",
|
229 |
+
"70": "NON_PARTICIPATING",
|
230 |
+
"71": "NON_PARTICIPATING",
|
231 |
+
"72": "NON_PARTICIPATING",
|
232 |
"73": "NON_PARTICIPATING",
|
233 |
"74": "NON_PARTICIPATING",
|
234 |
+
"75": "SUCCESS",
|
235 |
"76": "NON_PARTICIPATING",
|
236 |
"77": "NON_PARTICIPATING",
|
237 |
+
"78": "SUCCESS",
|
238 |
+
"79": "NON_PARTICIPATING",
|
239 |
"8": "NON_PARTICIPATING",
|
240 |
+
"80": "NON_PARTICIPATING",
|
241 |
"81": "NON_PARTICIPATING",
|
242 |
+
"82": "SUCCESS",
|
243 |
"83": "SUCCESS",
|
244 |
"84": "NON_PARTICIPATING",
|
245 |
+
"85": "NON_PARTICIPATING",
|
246 |
+
"86": "NON_PARTICIPATING",
|
247 |
"87": "NON_PARTICIPATING",
|
248 |
+
"88": "SUCCESS",
|
249 |
"89": "SUCCESS",
|
250 |
"9": "NON_PARTICIPATING",
|
251 |
+
"90": "NON_PARTICIPATING",
|
252 |
+
"91": "SUCCESS",
|
253 |
"92": "NON_PARTICIPATING",
|
254 |
"93": "NON_PARTICIPATING",
|
255 |
"94": "SUCCESS",
|
|
|
257 |
"96": "NON_PARTICIPATING",
|
258 |
"97": "NON_PARTICIPATING",
|
259 |
"98": "NON_PARTICIPATING",
|
260 |
+
"99": "NON_PARTICIPATING"
|
261 |
},
|
262 |
"architectures": [
|
263 |
"GPTOptim"
|
model-00001-of-00002.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4975530824
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:be5bdcb23e505b6ba49dd6573dd89d2b1b8f263dfaadad43b97a7d0d02049086
|
3 |
size 4975530824
|
model-00002-of-00002.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3075942944
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:57ae2ba67c6b1199ff03e7ba0ef6203ec1f3d4e5157416c13489c25ea40b9b80
|
3 |
size 3075942944
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4675755870
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7340e82d11e6d61abc23337a5cf1f85c4ab461122c17724303c0d9d7f81a8af9
|
3 |
size 4675755870
|