layer_id int64 0 223 | name stringlengths 26 32 | D float64 0.03 0.21 | M int64 1.02k 4.1k | N int64 4.1k 14.3k | Q float64 1 4 | alpha float64 2.55 30.7 | alpha_weighted float64 -100.7 -4.79 | entropy float64 1.07 1.57 | has_esd bool 1
class | lambda_max float32 0 0.01 | layer_type stringclasses 1
value | log_alpha_norm float64 -100.63 -4.73 | log_norm float32 -1.66 -0.94 | log_spectral_norm float32 -3.31 -1.88 | matrix_rank int64 64 64 | norm float32 0.02 0.11 | num_evals int64 1.02k 4.1k | num_pl_spikes int64 5 64 | rank_loss int64 960 4.03k | rf int64 1 1 | sigma float64 0.35 9.03 | spectral_norm float32 0 0.01 | stable_rank float32 4.91 54.6 | status stringclasses 1
value | sv_max float64 0.02 0.11 | 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 | 0.097567 | 4,096 | 4,096 | 1 | 13.455432 | -37.866849 | 1.56443 | true | 0.001534 | dense | -37.866766 | -1.429151 | -2.814243 | 64 | 0.037226 | 4,096 | 64 | 4,032 | 1 | 1.556929 | 0.001534 | 24.271242 | success | 0.039163 | 0 | under-trained | 4,032 | 0.001534 | 0.00053 |
201 | model.layers.28.self_attn.q_proj | 0.067804 | 4,096 | 4,096 | 1 | 8.690762 | -24.135214 | 1.561632 | true | 0.001671 | dense | -24.132781 | -1.462055 | -2.777111 | 64 | 0.03451 | 4,096 | 64 | 4,032 | 1 | 0.961345 | 0.001671 | 20.656462 | success | 0.040874 | 0 | under-trained | 4,032 | 0.001671 | 0.000464 |
202 | model.layers.28.self_attn.v_proj | 0.076671 | 1,024 | 4,096 | 4 | 21.357595 | -69.448551 | 1.136971 | true | 0.00056 | dense | -69.02072 | -1.538136 | -3.251703 | 64 | 0.028964 | 1,024 | 20 | 960 | 1 | 4.552097 | 0.00056 | 51.709057 | success | 0.023667 | 0.000001 | under-trained | 960 | 0.00056 | 0.000467 |
203 | model.layers.29.mlp.down_proj | 0.174697 | 4,096 | 14,336 | 3.5 | 3.982349 | -9.512021 | 1.560833 | true | 0.004087 | dense | -9.366647 | -1.121441 | -2.388545 | 64 | 0.075607 | 4,096 | 5 | 4,032 | 1 | 1.333747 | 0.004087 | 18.497126 | success | 0.063933 | 0.000001 | 4,032 | 0.004087 | 0.001206 | |
204 | model.layers.29.mlp.gate_proj | 0.110376 | 4,096 | 14,336 | 3.5 | 4.602726 | -10.848719 | 1.560916 | true | 0.004395 | dense | -10.739705 | -1.061237 | -2.35702 | 64 | 0.086849 | 4,096 | 8 | 4,032 | 1 | 1.273756 | 0.004395 | 19.759821 | success | 0.066296 | 0.000001 | 4,032 | 0.004395 | 0.001395 | |
205 | model.layers.29.mlp.up_proj | 0.117209 | 4,096 | 14,336 | 3.5 | 4.878604 | -11.948773 | 1.561445 | true | 0.003555 | dense | -11.741224 | -1.090014 | -2.449219 | 64 | 0.081281 | 4,096 | 12 | 4,032 | 1 | 1.119657 | 0.003555 | 22.866827 | success | 0.05962 | 0.000001 | 4,032 | 0.003555 | 0.001235 | |
206 | model.layers.29.self_attn.k_proj | 0.077428 | 1,024 | 4,096 | 4 | 6.71173 | -19.3982 | 1.127362 | true | 0.001288 | dense | -19.169781 | -1.536191 | -2.890194 | 64 | 0.029094 | 1,024 | 64 | 960 | 1 | 0.713966 | 0.001288 | 22.594532 | success | 0.035884 | 0.000001 | under-trained | 960 | 0.001288 | 0.000369 |
207 | model.layers.29.self_attn.o_proj | 0.0927 | 4,096 | 4,096 | 1 | 11.705402 | -31.38998 | 1.561996 | true | 0.002081 | dense | -31.389846 | -1.37154 | -2.681666 | 64 | 0.042507 | 4,096 | 64 | 4,032 | 1 | 1.338175 | 0.002081 | 20.423292 | success | 0.045621 | 0 | under-trained | 4,032 | 0.002081 | 0.000593 |
208 | model.layers.29.self_attn.q_proj | 0.06169 | 4,096 | 4,096 | 1 | 7.948959 | -21.362541 | 1.557784 | true | 0.002054 | dense | -21.361654 | -1.463992 | -2.687464 | 64 | 0.034356 | 4,096 | 64 | 4,032 | 1 | 0.86862 | 0.002054 | 16.729059 | success | 0.045318 | 0 | under-trained | 4,032 | 0.002054 | 0.000451 |
209 | model.layers.29.self_attn.v_proj | 0.058828 | 1,024 | 4,096 | 4 | 24.171611 | -77.009869 | 1.136972 | true | 0.000652 | dense | -76.959004 | -1.505554 | -3.185963 | 64 | 0.031221 | 1,024 | 17 | 960 | 1 | 5.619941 | 0.000652 | 47.908161 | success | 0.025528 | 0.000001 | under-trained | 960 | 0.000652 | 0.000508 |
210 | model.layers.30.mlp.down_proj | 0.132176 | 4,096 | 14,336 | 3.5 | 3.446907 | -8.061023 | 1.557716 | true | 0.004585 | dense | -7.880584 | -1.125973 | -2.338625 | 64 | 0.074822 | 4,096 | 6 | 4,032 | 1 | 0.998946 | 0.004585 | 16.317465 | success | 0.067715 | 0.000001 | 4,032 | 0.004585 | 0.001158 | |
211 | model.layers.30.mlp.gate_proj | 0.108602 | 4,096 | 14,336 | 3.5 | 5.523198 | -12.654264 | 1.559638 | true | 0.005116 | dense | -12.633893 | -1.042727 | -2.291112 | 64 | 0.09063 | 4,096 | 12 | 4,032 | 1 | 1.305735 | 0.005116 | 17.716766 | success | 0.071523 | 0.000001 | 4,032 | 0.005116 | 0.001391 | |
212 | model.layers.30.mlp.up_proj | 0.136689 | 4,096 | 14,336 | 3.5 | 4.606794 | -10.939204 | 1.5597 | true | 0.004221 | dense | -10.797425 | -1.076881 | -2.374581 | 64 | 0.083776 | 4,096 | 10 | 4,032 | 1 | 1.140568 | 0.004221 | 19.847208 | success | 0.06497 | 0.000001 | 4,032 | 0.004221 | 0.001308 | |
213 | model.layers.30.self_attn.k_proj | 0.06174 | 1,024 | 4,096 | 4 | 7.277368 | -20.806148 | 1.128495 | true | 0.001383 | dense | -20.728674 | -1.524596 | -2.859021 | 64 | 0.029882 | 1,024 | 64 | 960 | 1 | 0.784671 | 0.001383 | 21.598597 | success | 0.037195 | 0.000001 | under-trained | 960 | 0.001383 | 0.000387 |
214 | model.layers.30.self_attn.o_proj | 0.127274 | 4,096 | 4,096 | 1 | 4.013937 | -10.695391 | 1.559696 | true | 0.002165 | dense | -10.531604 | -1.390642 | -2.664564 | 64 | 0.040678 | 4,096 | 8 | 4,032 | 1 | 1.065588 | 0.002165 | 18.789783 | success | 0.046528 | 0 | 4,032 | 0.002165 | 0.000634 | |
215 | model.layers.30.self_attn.q_proj | 0.078681 | 4,096 | 4,096 | 1 | 7.80563 | -20.244153 | 1.555531 | true | 0.00255 | dense | -20.243469 | -1.409045 | -2.593532 | 64 | 0.03899 | 4,096 | 64 | 4,032 | 1 | 0.850704 | 0.00255 | 15.29279 | success | 0.050493 | 0 | under-trained | 4,032 | 0.00255 | 0.000507 |
216 | model.layers.30.self_attn.v_proj | 0.088427 | 1,024 | 4,096 | 4 | 22.691435 | -73.210394 | 1.137023 | true | 0.000594 | dense | -72.865368 | -1.51168 | -3.226345 | 64 | 0.030784 | 1,024 | 19 | 960 | 1 | 4.976356 | 0.000594 | 51.839954 | success | 0.024368 | 0.000001 | under-trained | 960 | 0.000594 | 0.000497 |
217 | model.layers.31.mlp.down_proj | 0.211114 | 4,096 | 14,336 | 3.5 | 3.167784 | -7.184908 | 1.544066 | true | 0.005394 | dense | -6.866488 | -1.098212 | -2.268118 | 64 | 0.079761 | 4,096 | 10 | 4,032 | 1 | 0.685513 | 0.005394 | 14.787909 | success | 0.073441 | 0.000001 | 4,032 | 0.005394 | 0.001204 | |
218 | model.layers.31.mlp.gate_proj | 0.09052 | 4,096 | 14,336 | 3.5 | 4.625173 | -10.50815 | 1.559803 | true | 0.005346 | dense | -10.436979 | -1.012104 | -2.271947 | 64 | 0.097252 | 4,096 | 10 | 4,032 | 1 | 1.146381 | 0.005346 | 18.190453 | success | 0.073118 | 0.000001 | 4,032 | 0.005346 | 0.001479 | |
219 | model.layers.31.mlp.up_proj | 0.129257 | 4,096 | 14,336 | 3.5 | 5.00781 | -11.894489 | 1.561102 | true | 0.004215 | dense | -11.763711 | -1.047329 | -2.375188 | 64 | 0.089675 | 4,096 | 11 | 4,032 | 1 | 1.2084 | 0.004215 | 21.274475 | success | 0.064924 | 0.000001 | 4,032 | 0.004215 | 0.001375 | |
220 | model.layers.31.self_attn.k_proj | 0.036201 | 1,024 | 4,096 | 4 | 7.879704 | -23.809126 | 1.133314 | true | 0.000952 | dense | -23.701685 | -1.557212 | -3.021576 | 64 | 0.02772 | 1,024 | 52 | 960 | 1 | 0.954043 | 0.000952 | 29.131598 | success | 0.030847 | 0.000001 | under-trained | 960 | 0.000952 | 0.000381 |
221 | model.layers.31.self_attn.o_proj | 0.144043 | 4,096 | 4,096 | 1 | 11.16667 | -28.305311 | 1.555512 | true | 0.002919 | dense | -28.305302 | -1.367495 | -2.534803 | 64 | 0.042905 | 4,096 | 64 | 4,032 | 1 | 1.270834 | 0.002919 | 14.699698 | success | 0.054025 | 0 | under-trained | 4,032 | 0.002919 | 0.000587 |
222 | model.layers.31.self_attn.q_proj | 0.077791 | 4,096 | 4,096 | 1 | 8.190492 | -22.01782 | 1.559381 | true | 0.00205 | dense | -22.015165 | -1.4183 | -2.688217 | 64 | 0.038168 | 4,096 | 64 | 4,032 | 1 | 0.898811 | 0.00205 | 18.617321 | success | 0.045278 | 0 | under-trained | 4,032 | 0.00205 | 0.000505 |
223 | model.layers.31.self_attn.v_proj | 0.074278 | 1,024 | 4,096 | 4 | 15.33425 | -48.471702 | 1.136779 | true | 0.00069 | dense | -48.338322 | -1.51178 | -3.161009 | 64 | 0.030777 | 1,024 | 39 | 960 | 1 | 2.295317 | 0.00069 | 44.589176 | success | 0.026272 | 0.000001 | under-trained | 960 | 0.00069 | 0.000466 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.