layer_id
int64 0
223
| name
stringlengths 26
32
| D
float64 0.02
0.18
| M
int64 1.02k
4.1k
| N
int64 4.1k
14.3k
| Q
float64 1
4
| alpha
float64 3.57
40.4
| alpha_weighted
float64 -123.19
-7.38
| entropy
float64 1.1
1.57
| has_esd
bool 1
class | lambda_max
float32 0
0.01
| layer_type
stringclasses 1
value | log_alpha_norm
float64 -123.03
-6.97
| log_norm
float32 -1.93
-0.76
| log_spectral_norm
float32 -3.35
-2
| matrix_rank
int64 64
64
| norm
float32 0.01
0.17
| num_evals
int64 1.02k
4.1k
| num_pl_spikes
int64 7
64
| rank_loss
int64 960
4.03k
| rf
int64 1
1
| sigma
float64 0.59
11.9
| spectral_norm
float32 0
0.01
| stable_rank
float32 14
56.5
| status
stringclasses 1
value | sv_max
float64 0.02
0.1
| sv_min
float64 0
0
| warning
stringclasses 2
values | weak_rank_loss
int64 960
4.03k
| xmax
float64 0
0.01
| xmin
float64 0
0
|
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
200 | model.layers.28.self_attn.o_proj | 0.097805 | 4,096 | 4,096 | 1 | 13.500109 | -36.692384 | 1.566192 | true | 0.001915 | dense | -36.688367 | -1.222852 | -2.717932 | 64 | 0.059862 | 4,096 | 64 | 4,032 | 1 | 1.562514 | 0.001915 | 31.266609 | success | 0.043756 | 0 | under-trained | 4,032 | 0.001915 | 0.000857 |
201 | model.layers.28.self_attn.q_proj | 0.032679 | 4,096 | 4,096 | 1 | 8.820552 | -23.808232 | 1.564349 | true | 0.001999 | dense | -23.786964 | -1.270111 | -2.699177 | 64 | 0.05369 | 4,096 | 63 | 4,032 | 1 | 0.985297 | 0.001999 | 26.857561 | success | 0.044711 | 0 | under-trained | 4,032 | 0.001999 | 0.000729 |
202 | model.layers.28.self_attn.v_proj | 0.05884 | 1,024 | 4,096 | 4 | 26.562079 | -80.719735 | 1.137033 | true | 0.000914 | dense | -80.605992 | -1.336471 | -3.038909 | 64 | 0.046082 | 1,024 | 15 | 960 | 1 | 6.600101 | 0.000914 | 50.400822 | success | 0.030237 | 0.000001 | under-trained | 960 | 0.000914 | 0.000756 |
203 | model.layers.29.mlp.down_proj | 0.116928 | 4,096 | 14,336 | 3.5 | 5.635298 | -12.798872 | 1.563257 | true | 0.005356 | dense | -12.7392 | -0.926453 | -2.271197 | 64 | 0.118453 | 4,096 | 8 | 4,032 | 1 | 1.638825 | 0.005356 | 22.117905 | success | 0.073182 | 0.000001 | 4,032 | 0.005356 | 0.001893 |
|
204 | model.layers.29.mlp.gate_proj | 0.095314 | 4,096 | 14,336 | 3.5 | 5.709415 | -11.900639 | 1.558272 | true | 0.008234 | dense | -11.889177 | -0.86286 | -2.084388 | 64 | 0.137132 | 4,096 | 14 | 4,032 | 1 | 1.258644 | 0.008234 | 16.654388 | success | 0.090741 | 0.000001 | 4,032 | 0.008234 | 0.002101 |
|
205 | model.layers.29.mlp.up_proj | 0.089241 | 4,096 | 14,336 | 3.5 | 5.499295 | -12.090405 | 1.560934 | true | 0.006331 | dense | -12.041353 | -0.894798 | -2.198537 | 64 | 0.12741 | 4,096 | 15 | 4,032 | 1 | 1.161713 | 0.006331 | 20.125153 | success | 0.079567 | 0.000001 | 4,032 | 0.006331 | 0.001928 |
|
206 | model.layers.29.self_attn.k_proj | 0.047418 | 1,024 | 4,096 | 4 | 8.436308 | -25.361052 | 1.135113 | true | 0.000986 | dense | -24.77831 | -1.396343 | -3.006179 | 64 | 0.040147 | 1,024 | 57 | 960 | 1 | 0.984963 | 0.000986 | 40.722656 | success | 0.031399 | 0.000001 | under-trained | 960 | 0.000986 | 0.000552 |
207 | model.layers.29.self_attn.o_proj | 0.112862 | 4,096 | 4,096 | 1 | 12.550207 | -31.941633 | 1.563841 | true | 0.00285 | dense | -31.941213 | -1.169205 | -2.545108 | 64 | 0.067732 | 4,096 | 64 | 4,032 | 1 | 1.443776 | 0.00285 | 23.763067 | success | 0.053388 | 0 | under-trained | 4,032 | 0.00285 | 0.000957 |
208 | model.layers.29.self_attn.q_proj | 0.031023 | 4,096 | 4,096 | 1 | 8.850884 | -24.35266 | 1.564856 | true | 0.001772 | dense | -24.327975 | -1.301656 | -2.751438 | 64 | 0.049928 | 4,096 | 61 | 4,032 | 1 | 1.005203 | 0.001772 | 28.169666 | success | 0.0421 | 0 | under-trained | 4,032 | 0.001772 | 0.000683 |
209 | model.layers.29.self_attn.v_proj | 0.082863 | 1,024 | 4,096 | 4 | 25.604042 | -76.527974 | 1.137019 | true | 0.001026 | dense | -76.441977 | -1.29374 | -2.988902 | 64 | 0.050846 | 1,024 | 16 | 960 | 1 | 6.15101 | 0.001026 | 49.56353 | success | 0.032029 | 0.000001 | under-trained | 960 | 0.001026 | 0.000831 |
210 | model.layers.30.mlp.down_proj | 0.121525 | 4,096 | 14,336 | 3.5 | 4.095597 | -8.869663 | 1.558633 | true | 0.006829 | dense | -8.754653 | -0.922824 | -2.165658 | 64 | 0.119447 | 4,096 | 7 | 4,032 | 1 | 1.170026 | 0.006829 | 17.491791 | success | 0.082636 | 0.000001 | 4,032 | 0.006829 | 0.00197 |
|
211 | model.layers.30.mlp.gate_proj | 0.095954 | 4,096 | 14,336 | 3.5 | 5.390292 | -10.79068 | 1.555076 | true | 0.009957 | dense | -10.781551 | -0.834772 | -2.001873 | 64 | 0.146295 | 4,096 | 16 | 4,032 | 1 | 1.097573 | 0.009957 | 14.692684 | success | 0.099785 | 0.000001 | 4,032 | 0.009957 | 0.002176 |
|
212 | model.layers.30.mlp.up_proj | 0.094755 | 4,096 | 14,336 | 3.5 | 4.092898 | -8.543034 | 1.556608 | true | 0.008179 | dense | -8.448714 | -0.872864 | -2.087282 | 64 | 0.13401 | 4,096 | 9 | 4,032 | 1 | 1.030966 | 0.008179 | 16.383945 | success | 0.09044 | 0.000001 | 4,032 | 0.008179 | 0.002164 |
|
213 | model.layers.30.self_attn.k_proj | 0.033068 | 1,024 | 4,096 | 4 | 8.648706 | -25.553527 | 1.134789 | true | 0.00111 | dense | -25.127229 | -1.372152 | -2.954607 | 64 | 0.042447 | 1,024 | 41 | 960 | 1 | 1.194527 | 0.00111 | 38.234444 | success | 0.033319 | 0.000001 | under-trained | 960 | 0.00111 | 0.000615 |
214 | model.layers.30.self_attn.o_proj | 0.148381 | 4,096 | 4,096 | 1 | 3.666627 | -9.161233 | 1.558121 | true | 0.003173 | dense | -8.763318 | -1.17302 | -2.498545 | 64 | 0.06714 | 4,096 | 9 | 4,032 | 1 | 0.888876 | 0.003173 | 21.160452 | success | 0.056328 | 0 | 4,032 | 0.003173 | 0.001034 |
|
215 | model.layers.30.self_attn.q_proj | 0.028241 | 4,096 | 4,096 | 1 | 8.020482 | -20.355868 | 1.561349 | true | 0.002897 | dense | -20.352326 | -1.227622 | -2.537986 | 64 | 0.059208 | 4,096 | 63 | 4,032 | 1 | 0.884498 | 0.002897 | 20.434505 | success | 0.053828 | 0 | under-trained | 4,032 | 0.002897 | 0.000786 |
216 | model.layers.30.self_attn.v_proj | 0.09179 | 1,024 | 4,096 | 4 | 32.460629 | -97.711113 | 1.137061 | true | 0.000977 | dense | -97.642366 | -1.301774 | -3.010142 | 64 | 0.049914 | 1,024 | 10 | 960 | 1 | 9.948724 | 0.000977 | 51.093761 | success | 0.031256 | 0.000001 | under-trained | 960 | 0.000977 | 0.000839 |
217 | model.layers.31.mlp.down_proj | 0.125934 | 4,096 | 14,336 | 3.5 | 3.571153 | -7.375295 | 1.537734 | true | 0.008605 | dense | -6.966406 | -0.87753 | -2.065242 | 64 | 0.132578 | 4,096 | 19 | 4,032 | 1 | 0.589863 | 0.008605 | 15.406788 | success | 0.092764 | 0.000001 | 4,032 | 0.008605 | 0.001777 |
|
218 | model.layers.31.mlp.gate_proj | 0.091779 | 4,096 | 14,336 | 3.5 | 4.415988 | -8.811718 | 1.555982 | true | 0.010106 | dense | -8.763337 | -0.804311 | -1.995413 | 64 | 0.156924 | 4,096 | 10 | 4,032 | 1 | 1.08023 | 0.010106 | 15.527523 | success | 0.10053 | 0.000001 | 4,032 | 0.010106 | 0.002479 |
|
219 | model.layers.31.mlp.up_proj | 0.086961 | 4,096 | 14,336 | 3.5 | 4.211193 | -8.774598 | 1.557385 | true | 0.008248 | dense | -8.662626 | -0.843593 | -2.083637 | 64 | 0.143353 | 4,096 | 10 | 4,032 | 1 | 1.015468 | 0.008248 | 17.379765 | success | 0.09082 | 0.000001 | 4,032 | 0.008248 | 0.00225 |
|
220 | model.layers.31.self_attn.k_proj | 0.028935 | 1,024 | 4,096 | 4 | 8.169591 | -23.565732 | 1.134226 | true | 0.001304 | dense | -23.368861 | -1.361035 | -2.884567 | 64 | 0.043548 | 1,024 | 40 | 960 | 1 | 1.133612 | 0.001304 | 33.383484 | success | 0.036117 | 0.000001 | under-trained | 960 | 0.001304 | 0.000628 |
221 | model.layers.31.self_attn.o_proj | 0.135698 | 4,096 | 4,096 | 1 | 10.803733 | -25.741085 | 1.557611 | true | 0.004144 | dense | -25.740958 | -1.164674 | -2.38261 | 64 | 0.068442 | 4,096 | 64 | 4,032 | 1 | 1.225467 | 0.004144 | 16.517179 | success | 0.064372 | 0 | under-trained | 4,032 | 0.004144 | 0.000936 |
222 | model.layers.31.self_attn.q_proj | 0.079925 | 4,096 | 4,096 | 1 | 7.315148 | -18.390798 | 1.559154 | true | 0.003061 | dense | -18.376024 | -1.221106 | -2.514071 | 64 | 0.060103 | 4,096 | 64 | 4,032 | 1 | 0.789393 | 0.003061 | 19.63199 | success | 0.055331 | 0 | under-trained | 4,032 | 0.003061 | 0.000778 |
223 | model.layers.31.self_attn.v_proj | 0.070425 | 1,024 | 4,096 | 4 | 17.833163 | -53.131643 | 1.136837 | true | 0.001049 | dense | -52.768134 | -1.29625 | -2.979373 | 64 | 0.050553 | 1,024 | 24 | 960 | 1 | 3.436055 | 0.001049 | 48.208412 | success | 0.032383 | 0.000001 | under-trained | 960 | 0.001049 | 0.000799 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.