layer_type | N | M | Q | alpha | D | alpha-hat | log_SN | % Rand | num_traps | num_fingers | rank_loss | |
---|---|---|---|---|---|---|---|---|---|---|---|---|
layer_id | ||||||||||||
3 | EMBEDDING | 49408 | 512 | 96.50 | 5.54 | 0.05 | 10.08 | 1.82 | 89.34 | 0 | 0 | 0 |
4 | EMBEDDING | 512 | 77 | 6.65 | 1.28 | 0.12 | -0.15 | -0.11 | 45.41 | 18 | 1 | 0 |
9 | DENSE | 512 | 512 | 1.00 | 8.60 | 0.11 | 1.49 | 0.17 | 89.38 | 0 | 3 | 4 |
10 | DENSE | 512 | 512 | 1.00 | 15.63 | 0.11 | -11.24 | -0.72 | 86.79 | 0 | 1 | 1 |
11 | DENSE | 512 | 512 | 1.00 | 13.13 | 0.07 | 2.22 | 0.17 | 91.88 | 0 | 0 | 3 |
12 | DENSE | 512 | 512 | 1.00 | 4.80 | 0.05 | -2.15 | -0.45 | 88.87 | 0 | 0 | 2 |
16 | DENSE | 2048 | 512 | 4.00 | 3.47 | 0.06 | 7.13 | 2.06 | 67.37 | 15 | 0 | 0 |
17 | DENSE | 2048 | 512 | 4.00 | 5.66 | 0.03 | 0.29 | 0.05 | 94.27 | 0 | 0 | 0 |
21 | DENSE | 512 | 512 | 1.00 | 4.02 | 0.13 | 1.39 | 0.35 | 90.83 | 0 | 6 | 5 |
22 | DENSE | 512 | 512 | 1.00 | 5.52 | 0.07 | -2.03 | -0.37 | 95.17 | 0 | 0 | 2 |
23 | DENSE | 512 | 512 | 1.00 | 3.82 | 0.12 | 1.34 | 0.35 | 91.10 | 0 | 4 | 4 |
24 | DENSE | 512 | 512 | 1.00 | 4.72 | 0.09 | -1.87 | -0.40 | 94.35 | 0 | 0 | 2 |
28 | DENSE | 2048 | 512 | 4.00 | 4.78 | 0.03 | 4.57 | 0.96 | 90.67 | 0 | 0 | 0 |
29 | DENSE | 2048 | 512 | 4.00 | 4.78 | 0.06 | 0.19 | 0.04 | 93.15 | 0 | 0 | 0 |
33 | DENSE | 512 | 512 | 1.00 | 4.77 | 0.06 | 1.08 | 0.23 | 93.62 | 0 | 0 | 4 |
34 | DENSE | 512 | 512 | 1.00 | 5.11 | 0.13 | -1.94 | -0.38 | 97.11 | 0 | 4 | 1 |
35 | DENSE | 512 | 512 | 1.00 | 5.10 | 0.06 | 1.70 | 0.33 | 90.92 | 0 | 0 | 4 |
36 | DENSE | 512 | 512 | 1.00 | 4.38 | 0.12 | -1.59 | -0.36 | 96.25 | 0 | 4 | 2 |
40 | DENSE | 2048 | 512 | 4.00 | 4.98 | 0.03 | 4.90 | 0.98 | 90.65 | 0 | 0 | 0 |
41 | DENSE | 2048 | 512 | 4.00 | 4.41 | 0.07 | 0.02 | 0.00 | 93.66 | 0 | 0 | 0 |
45 | DENSE | 512 | 512 | 1.00 | 3.66 | 0.06 | 1.02 | 0.28 | 90.39 | 0 | 0 | 3 |
46 | DENSE | 512 | 512 | 1.00 | 4.37 | 0.13 | -1.53 | -0.35 | 97.56 | 0 | 7 | 1 |
47 | DENSE | 512 | 512 | 1.00 | 2.92 | 0.10 | 1.03 | 0.35 | 87.31 | 0 | 1 | 2 |
48 | DENSE | 512 | 512 | 1.00 | 4.38 | 0.13 | -1.75 | -0.40 | 97.77 | 0 | 7 | 1 |
52 | DENSE | 2048 | 512 | 4.00 | 5.17 | 0.04 | 5.23 | 1.01 | 90.15 | 0 | 0 | 0 |
53 | DENSE | 2048 | 512 | 4.00 | 5.18 | 0.09 | -0.30 | -0.06 | 94.96 | 0 | 1 | 0 |
57 | DENSE | 512 | 512 | 1.00 | 4.88 | 0.07 | 1.10 | 0.23 | 93.19 | 0 | 0 | 0 |
58 | DENSE | 512 | 512 | 1.00 | 5.04 | 0.10 | -1.70 | -0.34 | 97.44 | 0 | 1 | 2 |
59 | DENSE | 512 | 512 | 1.00 | 3.14 | 0.12 | 0.89 | 0.28 | 88.66 | 0 | 5 | 1 |
60 | DENSE | 512 | 512 | 1.00 | 4.66 | 0.12 | -1.90 | -0.41 | 97.05 | 0 | 1 | 1 |
64 | DENSE | 2048 | 512 | 4.00 | 5.32 | 0.05 | 5.33 | 1.00 | 90.22 | 0 | 0 | 0 |
65 | DENSE | 2048 | 512 | 4.00 | 5.91 | 0.08 | -0.21 | -0.04 | 95.07 | 0 | 0 | 0 |
69 | DENSE | 512 | 512 | 1.00 | 4.78 | 0.07 | 0.54 | 0.11 | 94.39 | 0 | 0 | 2 |
70 | DENSE | 512 | 512 | 1.00 | 6.17 | 0.12 | -2.41 | -0.39 | 99.19 | 0 | 1 | 2 |
71 | DENSE | 512 | 512 | 1.00 | 3.60 | 0.13 | 1.01 | 0.28 | 90.36 | 0 | 6 | 2 |
72 | DENSE | 512 | 512 | 1.00 | 5.20 | 0.11 | -1.82 | -0.35 | 96.74 | 0 | 1 | 2 |
76 | DENSE | 2048 | 512 | 4.00 | 5.53 | 0.07 | 5.69 | 1.03 | 89.89 | 0 | 0 | 0 |
77 | DENSE | 2048 | 512 | 4.00 | 6.74 | 0.10 | 0.06 | 0.01 | 95.42 | 0 | 1 | 0 |
81 | DENSE | 512 | 512 | 1.00 | 5.48 | 0.06 | 0.92 | 0.17 | 93.21 | 0 | 0 | 1 |
82 | DENSE | 512 | 512 | 1.00 | 7.38 | 0.10 | -2.82 | -0.38 | 97.53 | 0 | 0 | 2 |
83 | DENSE | 512 | 512 | 1.00 | 2.64 | 0.12 | 0.80 | 0.30 | 87.83 | 0 | 3 | 3 |
84 | DENSE | 512 | 512 | 1.00 | 4.29 | 0.13 | -1.49 | -0.35 | 96.20 | 0 | 1 | 2 |
88 | DENSE | 2048 | 512 | 4.00 | 5.61 | 0.05 | 5.45 | 0.97 | 90.52 | 0 | 0 | 0 |
89 | DENSE | 2048 | 512 | 4.00 | 7.89 | 0.10 | 0.40 | 0.05 | 96.15 | 0 | 1 | 0 |
93 | DENSE | 512 | 512 | 1.00 | 5.43 | 0.05 | 0.73 | 0.13 | 94.68 | 0 | 0 | 2 |
94 | DENSE | 512 | 512 | 1.00 | 4.48 | 0.14 | -1.60 | -0.36 | 98.40 | 0 | 6 | 1 |
95 | DENSE | 512 | 512 | 1.00 | 5.21 | 0.04 | 1.69 | 0.32 | 91.79 | 0 | 0 | 1 |
96 | DENSE | 512 | 512 | 1.00 | 8.53 | 0.09 | -3.25 | -0.38 | 98.82 | 0 | 0 | 2 |
100 | DENSE | 2048 | 512 | 4.00 | 5.22 | 0.08 | 4.89 | 0.94 | 90.84 | 0 | 0 | 0 |
101 | DENSE | 2048 | 512 | 4.00 | 6.93 | 0.12 | 0.82 | 0.12 | 96.24 | 0 | 2 | 0 |
105 | DENSE | 512 | 512 | 1.00 | 4.44 | 0.07 | 0.49 | 0.11 | 94.41 | 0 | 0 | 1 |
106 | DENSE | 512 | 512 | 1.00 | 5.29 | 0.13 | -1.93 | -0.36 | 98.31 | 0 | 2 | 1 |
107 | DENSE | 512 | 512 | 1.00 | 5.38 | 0.05 | 2.11 | 0.39 | 92.15 | 0 | 0 | 2 |
108 | DENSE | 512 | 512 | 1.00 | 7.31 | 0.12 | -2.94 | -0.40 | 98.61 | 0 | 1 | 1 |
112 | DENSE | 2048 | 512 | 4.00 | 5.85 | 0.06 | 4.99 | 0.85 | 91.66 | 0 | 0 | 0 |
113 | DENSE | 2048 | 512 | 4.00 | 6.32 | 0.13 | 1.56 | 0.25 | 95.90 | 0 | 6 | 0 |
117 | DENSE | 512 | 512 | 1.00 | 5.37 | 0.06 | 0.40 | 0.07 | 94.66 | 0 | 0 | 1 |
118 | DENSE | 512 | 512 | 1.00 | 7.17 | 0.08 | -1.64 | -0.23 | 98.23 | 0 | 0 | 1 |
119 | DENSE | 512 | 512 | 1.00 | 5.19 | 0.04 | 1.69 | 0.33 | 92.52 | 0 | 0 | 1 |
120 | DENSE | 512 | 512 | 1.00 | 8.71 | 0.05 | -0.92 | -0.11 | 97.02 | 0 | 0 | 1 |
124 | DENSE | 2048 | 512 | 4.00 | 5.78 | 0.05 | 4.55 | 0.79 | 92.12 | 0 | 0 | 0 |
125 | DENSE | 2048 | 512 | 4.00 | 8.25 | 0.04 | 2.58 | 0.31 | 94.71 | 0 | 0 | 0 |
129 | DENSE | 512 | 512 | 1.00 | 2.79 | 0.09 | 0.33 | 0.12 | 90.22 | 0 | 1 | 1 |
130 | DENSE | 512 | 512 | 1.00 | 4.53 | 0.14 | -1.22 | -0.27 | 97.42 | 0 | 0 | 1 |
131 | DENSE | 512 | 512 | 1.00 | 3.26 | 0.11 | 1.38 | 0.42 | 87.52 | 0 | 1 | 1 |
132 | DENSE | 512 | 512 | 1.00 | 12.53 | 0.12 | -4.06 | -0.32 | 95.82 | 0 | 0 | 2 |
136 | DENSE | 2048 | 512 | 4.00 | 5.57 | 0.05 | 3.82 | 0.69 | 91.92 | 0 | 0 | 0 |
137 | DENSE | 2048 | 512 | 4.00 | 4.17 | 0.11 | 1.60 | 0.38 | 94.16 | 0 | 3 | 0 |
141 | DENSE | 512 | 512 | 1.00 | 3.84 | 0.07 | 0.28 | 0.07 | 88.66 | 0 | 0 | 2 |
142 | DENSE | 512 | 512 | 1.00 | 3.07 | 0.11 | 0.01 | 0.00 | 98.09 | 0 | 1 | 2 |
143 | DENSE | 512 | 512 | 1.00 | 3.46 | 0.06 | 0.29 | 0.08 | 88.77 | 0 | 0 | 2 |
144 | DENSE | 512 | 512 | 1.00 | 7.62 | 0.11 | -0.72 | -0.09 | 94.79 | 0 | 1 | 2 |
148 | DENSE | 2048 | 512 | 4.00 | 4.40 | 0.04 | 3.11 | 0.71 | 91.23 | 0 | 0 | 0 |
149 | DENSE | 2048 | 512 | 4.00 | 4.62 | 0.03 | 3.09 | 0.67 | 91.33 | 0 | 0 | 0 |
155 | EMBEDDING | 768 | 50 | 15.36 | 1.41 | 0.10 | 2.09 | 1.48 | 46.64 | 5 | 1 | 0 |
161 | DENSE | 768 | 768 | 1.00 | 3.59 | 0.04 | 4.40 | 1.23 | 68.57 | 0 | 0 | 96 |
162 | DENSE | 768 | 768 | 1.00 | 3.70 | 0.04 | 3.34 | 0.90 | 82.74 | 0 | 0 | 14 |
163 | DENSE | 768 | 768 | 1.00 | 3.52 | 0.04 | 4.92 | 1.40 | 67.25 | 0 | 0 | 96 |
164 | DENSE | 768 | 768 | 1.00 | 3.96 | 0.05 | 0.23 | 0.06 | 91.92 | 0 | 0 | 3 |
168 | DENSE | 3072 | 768 | 4.00 | 2.78 | 0.05 | 5.39 | 1.94 | 62.37 | 0 | 0 | 0 |
169 | DENSE | 3072 | 768 | 4.00 | 2.69 | 0.02 | 2.98 | 1.11 | 71.86 | 0 | 0 | 0 |
173 | DENSE | 768 | 768 | 1.00 | 2.53 | 0.04 | 1.98 | 0.78 | 79.16 | 0 | 0 | 27 |
174 | DENSE | 768 | 768 | 1.00 | 4.80 | 0.03 | -0.42 | -0.09 | 86.71 | 0 | 0 | 3 |
175 | DENSE | 768 | 768 | 1.00 | 2.59 | 0.04 | 2.09 | 0.80 | 78.80 | 0 | 0 | 26 |
176 | DENSE | 768 | 768 | 1.00 | 4.20 | 0.03 | 0.77 | 0.18 | 85.70 | 0 | 0 | 2 |
180 | DENSE | 3072 | 768 | 4.00 | 3.11 | 0.02 | 3.51 | 1.13 | 87.06 | 0 | 0 | 0 |
181 | DENSE | 3072 | 768 | 4.00 | 5.73 | 0.03 | 1.01 | 0.18 | 91.23 | 0 | 0 | 0 |
185 | DENSE | 768 | 768 | 1.00 | 3.18 | 0.07 | 2.01 | 0.63 | 73.55 | 0 | 0 | 25 |
186 | DENSE | 768 | 768 | 1.00 | 4.54 | 0.07 | 0.14 | 0.03 | 93.53 | 0 | 0 | 2 |
187 | DENSE | 768 | 768 | 1.00 | 2.56 | 0.08 | 1.46 | 0.57 | 74.14 | 0 | 0 | 24 |
188 | DENSE | 768 | 768 | 1.00 | 4.12 | 0.03 | 0.08 | 0.02 | 92.51 | 0 | 0 | 4 |
192 | DENSE | 3072 | 768 | 4.00 | 5.75 | 0.04 | 3.60 | 0.63 | 87.37 | 0 | 0 | 0 |
193 | DENSE | 3072 | 768 | 4.00 | 4.91 | 0.05 | 1.20 | 0.25 | 89.93 | 0 | 0 | 0 |
197 | DENSE | 768 | 768 | 1.00 | 1.93 | 0.11 | 1.03 | 0.53 | 75.86 | 0 | 3 | 17 |
198 | DENSE | 768 | 768 | 1.00 | 4.50 | 0.04 | 0.34 | 0.08 | 90.69 | 0 | 0 | 2 |
199 | DENSE | 768 | 768 | 1.00 | 3.32 | 0.08 | 1.48 | 0.45 | 81.22 | 0 | 0 | 19 |
200 | DENSE | 768 | 768 | 1.00 | 4.54 | 0.04 | -0.20 | -0.04 | 93.67 | 0 | 0 | 3 |
204 | DENSE | 3072 | 768 | 4.00 | 4.28 | 0.02 | 2.68 | 0.63 | 91.36 | 0 | 0 | 0 |
205 | DENSE | 3072 | 768 | 4.00 | 6.11 | 0.04 | 1.59 | 0.26 | 91.77 | 0 | 0 | 0 |
209 | DENSE | 768 | 768 | 1.00 | 3.45 | 0.06 | 1.52 | 0.44 | 82.84 | 0 | 0 | 10 |
210 | DENSE | 768 | 768 | 1.00 | 5.17 | 0.07 | -0.25 | -0.05 | 93.55 | 0 | 0 | 2 |
211 | DENSE | 768 | 768 | 1.00 | 2.34 | 0.07 | 1.05 | 0.45 | 79.02 | 0 | 0 | 11 |
212 | DENSE | 768 | 768 | 1.00 | 5.24 | 0.06 | -0.51 | -0.10 | 94.43 | 0 | 0 | 3 |
216 | DENSE | 3072 | 768 | 4.00 | 4.52 | 0.02 | 3.06 | 0.68 | 91.23 | 0 | 0 | 0 |
217 | DENSE | 3072 | 768 | 4.00 | 5.65 | 0.04 | 1.52 | 0.27 | 93.79 | 0 | 0 | 0 |
221 | DENSE | 768 | 768 | 1.00 | 4.78 | 0.07 | 1.76 | 0.37 | 87.21 | 0 | 0 | 3 |
222 | DENSE | 768 | 768 | 1.00 | 3.12 | 0.09 | -0.02 | -0.01 | 93.58 | 0 | 0 | 2 |
223 | DENSE | 768 | 768 | 1.00 | 2.59 | 0.08 | 1.12 | 0.43 | 82.58 | 0 | 0 | 2 |
224 | DENSE | 768 | 768 | 1.00 | 6.04 | 0.08 | -0.62 | -0.10 | 94.88 | 0 | 0 | 1 |
228 | DENSE | 3072 | 768 | 4.00 | 4.78 | 0.02 | 3.90 | 0.81 | 91.16 | 0 | 0 | 0 |
229 | DENSE | 3072 | 768 | 4.00 | 4.80 | 0.06 | 1.50 | 0.31 | 94.21 | 0 | 0 | 0 |
233 | DENSE | 768 | 768 | 1.00 | 4.46 | 0.06 | 1.91 | 0.43 | 89.34 | 0 | 0 | 2 |
234 | DENSE | 768 | 768 | 1.00 | 4.10 | 0.09 | -0.23 | -0.06 | 91.68 | 0 | 0 | 2 |
235 | DENSE | 768 | 768 | 1.00 | 3.30 | 0.09 | 1.28 | 0.39 | 87.15 | 0 | 1 | 2 |
236 | DENSE | 768 | 768 | 1.00 | 4.04 | 0.10 | -0.24 | -0.06 | 94.17 | 0 | 3 | 2 |
240 | DENSE | 3072 | 768 | 4.00 | 4.87 | 0.02 | 5.04 | 1.04 | 90.81 | 0 | 0 | 0 |
241 | DENSE | 3072 | 768 | 4.00 | 5.37 | 0.06 | 2.03 | 0.38 | 93.95 | 0 | 0 | 0 |
245 | DENSE | 768 | 768 | 1.00 | 3.32 | 0.07 | 1.61 | 0.48 | 87.80 | 0 | 0 | 2 |
246 | DENSE | 768 | 768 | 1.00 | 4.46 | 0.10 | -0.16 | -0.04 | 93.72 | 0 | 1 | 2 |
247 | DENSE | 768 | 768 | 1.00 | 2.82 | 0.09 | 1.05 | 0.37 | 87.38 | 0 | 0 | 2 |
248 | DENSE | 768 | 768 | 1.00 | 4.92 | 0.04 | 0.11 | 0.02 | 94.88 | 0 | 0 | 3 |
252 | DENSE | 3072 | 768 | 4.00 | 4.94 | 0.03 | 5.46 | 1.10 | 90.69 | 0 | 0 | 0 |
253 | DENSE | 3072 | 768 | 4.00 | 5.99 | 0.06 | 2.18 | 0.36 | 95.77 | 0 | 0 | 0 |
257 | DENSE | 768 | 768 | 1.00 | 4.09 | 0.05 | 1.53 | 0.37 | 89.75 | 0 | 0 | 2 |
258 | DENSE | 768 | 768 | 1.00 | 9.17 | 0.10 | -1.35 | -0.15 | 94.69 | 0 | 0 | 2 |
259 | DENSE | 768 | 768 | 1.00 | 3.16 | 0.10 | 1.21 | 0.38 | 87.71 | 0 | 0 | 1 |
260 | DENSE | 768 | 768 | 1.00 | 6.23 | 0.04 | 0.60 | 0.10 | 96.24 | 0 | 0 | 2 |
264 | DENSE | 3072 | 768 | 4.00 | 5.24 | 0.03 | 6.24 | 1.19 | 89.97 | 0 | 0 | 0 |
265 | DENSE | 3072 | 768 | 4.00 | 6.72 | 0.10 | 3.57 | 0.53 | 96.14 | 0 | 6 | 0 |
269 | DENSE | 768 | 768 | 1.00 | 5.22 | 0.05 | 1.67 | 0.32 | 93.47 | 0 | 0 | 1 |
270 | DENSE | 768 | 768 | 1.00 | 6.91 | 0.12 | -0.80 | -0.12 | 97.65 | 0 | 1 | 2 |
271 | DENSE | 768 | 768 | 1.00 | 3.81 | 0.10 | 1.10 | 0.29 | 91.99 | 0 | 1 | 2 |
272 | DENSE | 768 | 768 | 1.00 | 6.07 | 0.11 | 0.72 | 0.12 | 96.57 | 0 | 7 | 2 |
276 | DENSE | 3072 | 768 | 4.00 | 5.09 | 0.05 | 7.60 | 1.49 | 86.94 | 0 | 0 | 0 |
277 | DENSE | 3072 | 768 | 4.00 | 7.23 | 0.10 | 3.44 | 0.48 | 96.23 | 0 | 1 | 0 |
281 | DENSE | 768 | 768 | 1.00 | 7.06 | 0.04 | 2.32 | 0.33 | 96.63 | 0 | 0 | 1 |
282 | DENSE | 768 | 768 | 1.00 | 6.61 | 0.12 | 1.10 | 0.17 | 97.12 | 0 | 4 | 1 |
283 | DENSE | 768 | 768 | 1.00 | 4.59 | 0.08 | 1.19 | 0.26 | 94.64 | 0 | 0 | 1 |
284 | DENSE | 768 | 768 | 1.00 | 5.43 | 0.05 | 0.28 | 0.05 | 96.37 | 0 | 0 | 1 |
288 | DENSE | 3072 | 768 | 4.00 | 5.61 | 0.07 | 7.51 | 1.34 | 89.64 | 0 | 0 | 0 |
289 | DENSE | 3072 | 768 | 4.00 | 6.29 | 0.03 | 4.37 | 0.70 | 92.24 | 0 | 0 | 0 |
293 | DENSE | 768 | 768 | 1.00 | 3.65 | 0.06 | 0.94 | 0.26 | 90.07 | 0 | 0 | 0 |
294 | DENSE | 768 | 768 | 1.00 | 9.94 | 0.11 | 0.62 | 0.06 | 97.30 | 0 | 2 | 1 |
295 | DENSE | 768 | 768 | 1.00 | 3.61 | 0.08 | 1.21 | 0.34 | 88.13 | 0 | 0 | 2 |
296 | DENSE | 768 | 768 | 1.00 | 7.10 | 0.08 | 3.17 | 0.45 | 92.58 | 0 | 0 | 3 |
300 | DENSE | 3072 | 768 | 4.00 | 4.33 | 0.03 | 4.88 | 1.13 | 87.93 | 0 | 0 | 0 |
301 | DENSE | 3072 | 768 | 4.00 | 3.42 | 0.04 | 2.65 | 0.78 | 74.18 | 0 | 0 | 0 |
304 | DENSE | 768 | 512 | 1.50 | 4.74 | 0.04 | -0.14 | -0.03 | 82.34 | 0 | 0 | 1 |
305 | DENSE | 512 | 512 | 1.00 | 7.70 | 0.06 | -1.60 | -0.21 | 73.78 | 0 | 0 | 1 |