Llama-Guard-3-8B


Find this model in the Llama-Guard model summary

Llama-Guard-3-8B Model Set Plots

Llama-Guard Compared to Base Model Plots

Llama-Guard-3-8B Model Selected Details
id layer_type N M Q alpha D alpha-hat num_spikes warning
1 dense 14336 4096 3.5 2.895462 0.015540 -7.623660 219
2 dense 14336 4096 3.5 2.933240 0.016520 -7.118763 94
3 dense 14336 4096 3.5 2.893848 0.021436 -6.932288 104
4 dense 4096 1024 4.0 1.888594 0.050264 -5.514302 65 over-trained
5 dense 4096 4096 1.0 2.121145 0.020482 -4.723256 383
6 dense 4096 4096 1.0 2.070021 0.011280 -3.993863 272
7 dense 4096 1024 4.0 1.980044 0.036726 -5.361856 43 over-trained
8 dense 4096 1024 4.0 2.488606 0.023898 -7.295724 196
9 dense 4096 4096 1.0 2.785572 0.058191 -7.130071 19
10 dense 4096 1024 4.0 1.948088 0.090339 -5.883610 64 over-trained
11 dense 4096 4096 1.0 2.330968 0.013882 -5.213944 222
12 dense 14336 4096 3.5 2.958781 0.016381 -7.053380 123
13 dense 14336 4096 3.5 2.762901 0.015146 -7.025739 169
14 dense 14336 4096 3.5 2.890910 0.018185 -6.822055 131
15 dense 14336 4096 3.5 2.616601 0.018445 -6.543152 141
16 dense 14336 4096 3.5 2.830448 0.014182 -6.637420 275
17 dense 14336 4096 3.5 2.744731 0.015059 -6.391057 290
18 dense 4096 1024 4.0 2.351740 0.046077 -7.614999 84
19 dense 4096 4096 1.0 2.193380 0.017622 -5.458714 309
20 dense 4096 4096 1.0 2.269593 0.016077 -6.204188 657
21 dense 4096 1024 4.0 2.333374 0.017364 -6.747411 163
22 dense 4096 4096 1.0 2.282339 0.021729 -5.890393 650
23 dense 4096 1024 4.0 2.425425 0.013399 -6.889395 160
24 dense 4096 1024 4.0 2.362155 0.058175 -7.728102 115
25 dense 4096 4096 1.0 2.228788 0.017681 -5.359694 205
26 dense 14336 4096 3.5 2.654454 0.007591 -5.578584 339
27 dense 14336 4096 3.5 2.354029 0.017770 -5.510922 401
28 dense 14336 4096 3.5 2.494307 0.012267 -5.050215 352
29 dense 14336 4096 3.5 2.240821 0.010743 -5.036624 578
30 dense 14336 4096 3.5 2.448441 0.007180 -5.021391 570
31 dense 14336 4096 3.5 2.242231 0.013300 -4.472530 499
32 dense 4096 1024 4.0 2.244187 0.048887 -7.484798 93
33 dense 4096 4096 1.0 2.097750 0.021055 -4.884035 230
34 dense 4096 4096 1.0 2.224113 0.027535 -5.754789 583
35 dense 4096 1024 4.0 2.305025 0.011905 -6.632420 147
36 dense 4096 1024 4.0 2.318683 0.032391 -7.756092 131
37 dense 4096 1024 4.0 2.229827 0.020368 -6.445964 135
38 dense 4096 4096 1.0 2.207722 0.015490 -5.798491 643
39 dense 4096 4096 1.0 2.054641 0.023082 -5.065136 245
40 dense 14336 4096 3.5 2.335911 0.005780 -4.716007 745
41 dense 14336 4096 3.5 2.169485 0.011507 -4.776388 638
42 dense 14336 4096 3.5 2.138547 0.011344 -4.227294 779
43 dense 4096 4096 1.0 2.138361 0.019408 -5.535222 321
44 dense 14336 4096 3.5 2.084725 0.008741 -4.580347 957
45 dense 14336 4096 3.5 2.269227 0.004508 -4.554233 935
46 dense 14336 4096 3.5 2.087514 0.008639 -4.105149 900
47 dense 4096 1024 4.0 2.305449 0.034048 -7.721116 117
48 dense 4096 4096 1.0 1.977387 0.020671 -4.940266 510 over-trained
49 dense 4096 1024 4.0 2.221656 0.024837 -6.439673 142
50 dense 4096 4096 1.0 1.928802 0.016415 -4.728353 451 over-trained
51 dense 4096 4096 1.0 2.109381 0.015289 -5.375088 351
52 dense 4096 1024 4.0 2.297168 0.032153 -7.701541 134
53 dense 4096 1024 4.0 2.129783 0.025873 -6.182943 155
54 dense 14336 4096 3.5 2.220529 0.004686 -4.330837 1117
55 dense 14336 4096 3.5 2.059748 0.004119 -4.481433 1235
56 dense 14336 4096 3.5 2.063506 0.004836 -4.009523 1287
57 dense 14336 4096 3.5 2.036129 0.005789 -4.418326 1597
58 dense 14336 4096 3.5 2.207350 0.005439 -4.243062 1290
59 dense 14336 4096 3.5 2.055113 0.005343 -3.943116 1482
60 dense 4096 1024 4.0 2.270984 0.026554 -7.656709 117
61 dense 4096 4096 1.0 1.916896 0.014209 -4.669682 408 over-trained
62 dense 4096 4096 1.0 2.113332 0.016763 -5.365679 441
63 dense 4096 1024 4.0 2.104918 0.034533 -6.392847 156
64 dense 14336 4096 3.5 2.049430 0.007407 -3.933889 1682
65 dense 4096 1024 4.0 2.307826 0.019705 -7.662163 130
66 dense 4096 4096 1.0 1.923469 0.014298 -4.637833 390 over-trained
67 dense 4096 4096 1.0 2.086175 0.017835 -5.282777 353
68 dense 4096 1024 4.0 2.066406 0.035843 -6.166479 140
69 dense 14336 4096 3.5 2.033933 0.010494 -4.328748 1791
70 dense 14336 4096 3.5 2.191304 0.007691 -4.165606 1448
71 dense 4096 1024 4.0 2.053286 0.035188 -6.189205 127
72 dense 14336 4096 3.5 1.975317 0.012694 -4.284787 607 over-trained
73 dense 14336 4096 3.5 2.168851 0.010913 -4.177765 1595
74 dense 14336 4096 3.5 2.017240 0.011727 -3.938433 659
75 dense 4096 1024 4.0 2.258777 0.028031 -7.620941 121
76 dense 4096 4096 1.0 1.892056 0.018755 -4.743715 352 over-trained
77 dense 4096 4096 1.0 2.108703 0.015739 -5.379592 352
78 dense 14336 4096 3.5 1.946388 0.013762 -4.185643 487 over-trained
79 dense 14336 4096 3.5 2.157306 0.012567 -4.212000 1711
80 dense 14336 4096 3.5 1.999195 0.016038 -3.963250 509 over-trained
81 dense 4096 1024 4.0 2.247360 0.021743 -7.746437 122
82 dense 4096 4096 1.0 1.860967 0.017987 -4.585101 299 over-trained
83 dense 4096 4096 1.0 2.070628 0.017465 -5.295914 348
84 dense 4096 1024 4.0 1.964194 0.047902 -5.961183 118 over-trained
85 dense 4096 4096 1.0 2.021917 0.017080 -4.845223 267
86 dense 14336 4096 3.5 1.931455 0.013004 -4.153117 507 over-trained
87 dense 14336 4096 3.5 2.064291 0.013718 -3.967159 435
88 dense 14336 4096 3.5 1.981411 0.014956 -3.848415 488 over-trained
89 dense 4096 1024 4.0 2.161950 0.019713 -6.980143 100
90 dense 4096 1024 4.0 1.989381 0.044464 -6.228686 119 over-trained
91 dense 4096 4096 1.0 1.847130 0.022316 -4.772400 319 over-trained
92 dense 14336 4096 3.5 2.012961 0.015566 -3.762992 353
93 dense 14336 4096 3.5 1.917863 0.014539 -4.105924 532 over-trained
94 dense 4096 1024 4.0 1.954423 0.040512 -6.077313 109 over-trained
95 dense 14336 4096 3.5 1.951066 0.017399 -3.717304 451 over-trained
96 dense 4096 1024 4.0 2.273067 0.020719 -7.595114 130
97 dense 4096 4096 1.0 1.828944 0.021806 -4.492669 278 over-trained
98 dense 4096 4096 1.0 2.004060 0.021968 -4.604826 254
99 dense 4096 1024 4.0 2.180189 0.023651 -7.296459 98
100 dense 4096 4096 1.0 1.810075 0.023864 -4.376760 261 over-trained
101 dense 4096 4096 1.0 1.986507 0.023179 -4.606633 236 over-trained
102 dense 4096 1024 4.0 2.048601 0.052332 -6.393292 71
103 dense 14336 4096 3.5 2.006461 0.014786 -3.711650 384
104 dense 14336 4096 3.5 1.919567 0.016441 -3.620362 417 over-trained
105 dense 14336 4096 3.5 1.863658 0.015468 -3.948923 440 over-trained
106 dense 4096 4096 1.0 1.821968 0.022042 -4.383362 278 over-trained
107 dense 4096 4096 1.0 1.936736 0.019798 -4.389825 186 over-trained
108 dense 4096 1024 4.0 2.132113 0.032563 -7.129221 98
109 dense 14336 4096 3.5 1.868096 0.015895 -3.436274 409 over-trained
110 dense 14336 4096 3.5 1.933947 0.016116 -3.525038 325 over-trained
111 dense 4096 1024 4.0 2.025563 0.049720 -6.369242 64
112 dense 14336 4096 3.5 1.847674 0.016266 -3.810018 505 over-trained
113 dense 4096 1024 4.0 1.994786 0.046161 -5.956581 87 over-trained
114 dense 4096 1024 4.0 2.068061 0.035502 -6.588052 90
115 dense 14336 4096 3.5 1.807569 0.022589 -3.818233 475 over-trained
116 dense 14336 4096 3.5 1.921177 0.015406 -3.489429 365 over-trained
117 dense 14336 4096 3.5 1.846560 0.015540 -3.358353 399 over-trained
118 dense 4096 4096 1.0 1.792282 0.030203 -4.233013 249 over-trained
119 dense 4096 4096 1.0 1.899628 0.027463 -4.485989 187 over-trained
120 dense 14336 4096 3.5 1.815693 0.017716 -3.593526 434 over-trained
121 dense 14336 4096 3.5 1.881655 0.018965 -3.410735 359 over-trained
122 dense 4096 1024 4.0 1.932484 0.034337 -5.541707 103 over-trained
123 dense 4096 4096 1.0 1.868322 0.024489 -4.255665 161 over-trained
124 dense 4096 4096 1.0 1.808222 0.021163 -4.117835 255 over-trained
125 dense 4096 1024 4.0 2.046056 0.026360 -6.205271 72
126 dense 14336 4096 3.5 1.813573 0.016836 -3.246586 387 over-trained
127 dense 14336 4096 3.5 1.873215 0.015295 -3.368088 329 over-trained
128 dense 14336 4096 3.5 1.818033 0.017226 -3.538606 402 over-trained
129 dense 14336 4096 3.5 1.809042 0.014374 -3.165155 350 over-trained
130 dense 4096 1024 4.0 2.123704 0.042653 -6.053444 31
131 dense 4096 4096 1.0 1.862524 0.029630 -4.525778 185 over-trained
132 dense 4096 4096 1.0 1.793526 0.027267 -3.993862 225 over-trained
133 dense 4096 1024 4.0 2.055964 0.027780 -6.493327 81
134 dense 4096 1024 4.0 2.028341 0.033439 -6.327989 70
135 dense 14336 4096 3.5 1.878056 0.014077 -3.298087 321 over-trained
136 dense 4096 4096 1.0 1.843742 0.023245 -4.384901 159 over-trained
137 dense 4096 1024 4.0 1.865500 0.031120 -5.020645 89 over-trained
138 dense 14336 4096 3.5 1.811022 0.017639 -3.447236 375 over-trained
139 dense 14336 4096 3.5 1.816436 0.013140 -3.125107 368 over-trained
140 dense 4096 4096 1.0 1.806780 0.017434 -3.879408 213 over-trained
141 dense 14336 4096 3.5 1.819224 0.014532 -2.962640 367 over-trained
142 dense 14336 4096 3.5 1.799052 0.019858 -3.350633 359 over-trained
143 dense 4096 1024 4.0 1.848512 0.031380 -5.049456 78 over-trained
144 dense 4096 4096 1.0 1.857550 0.028418 -4.146016 175 over-trained
145 dense 4096 4096 1.0 1.777349 0.025053 -3.738582 195 over-trained
146 dense 4096 1024 4.0 1.990807 0.044605 -6.182109 75 over-trained
147 dense 14336 4096 3.5 1.879099 0.015099 -3.151116 312 over-trained
148 dense 14336 4096 3.5 1.791530 0.017502 -3.128761 339 over-trained
149 dense 14336 4096 3.5 1.873355 0.015856 -3.045295 315 over-trained
150 dense 14336 4096 3.5 1.809959 0.013987 -2.790861 345 over-trained
151 dense 4096 1024 4.0 1.970657 0.041428 -5.554743 55 over-trained
152 dense 4096 4096 1.0 1.782579 0.015935 -3.549075 179 over-trained
153 dense 4096 4096 1.0 1.840857 0.018251 -3.996516 130 over-trained
154 dense 4096 1024 4.0 1.809656 0.048952 -4.726841 76 over-trained
155 dense 14336 4096 3.5 1.806751 0.013296 -2.709699 319 over-trained
156 dense 4096 1024 4.0 1.949886 0.024096 -5.345003 67 over-trained
157 dense 4096 4096 1.0 1.885568 0.038668 -4.622312 204 over-trained
158 dense 4096 4096 1.0 1.760535 0.025338 -3.880808 192 over-trained
159 dense 14336 4096 3.5 1.794917 0.018741 -3.096594 339 over-trained
160 dense 14336 4096 3.5 1.870773 0.015064 -2.874374 311 over-trained
161 dense 4096 1024 4.0 2.014086 0.035216 -6.426813 64
162 dense 4096 4096 1.0 2.095483 0.034375 -5.066926 36
163 dense 14336 4096 3.5 1.788324 0.017601 -3.046431 338 over-trained
164 dense 14336 4096 3.5 1.866467 0.017420 -2.802124 303 over-trained
165 dense 14336 4096 3.5 1.801072 0.015708 -2.659143 298 over-trained
166 dense 4096 1024 4.0 1.979189 0.051918 -6.301839 59 over-trained
167 dense 4096 4096 1.0 1.761120 0.032655 -3.766144 171 over-trained
168 dense 4096 1024 4.0 1.879100 0.025823 -5.056408 62 over-trained
169 dense 4096 1024 4.0 1.912883 0.023979 -5.297983 76 over-trained
170 dense 4096 4096 1.0 1.851471 0.036366 -4.232883 179 over-trained
171 dense 4096 4096 1.0 1.728640 0.035757 -3.762669 169 over-trained
172 dense 14336 4096 3.5 1.782652 0.020626 -2.951861 312 over-trained
173 dense 14336 4096 3.5 1.802172 0.023148 -2.581460 300 over-trained
174 dense 14336 4096 3.5 1.875481 0.025215 -2.724459 307 over-trained
175 dense 4096 1024 4.0 1.918628 0.034323 -5.744888 57 over-trained
176 dense 14336 4096 3.5 1.766763 0.022392 -2.798228 273 over-trained
177 dense 14336 4096 3.5 1.868498 0.027218 -2.674806 286 over-trained
178 dense 14336 4096 3.5 1.801312 0.028831 -2.565973 297 over-trained
179 dense 4096 1024 4.0 1.950293 0.049401 -6.376083 62 over-trained
180 dense 4096 4096 1.0 1.745506 0.026920 -3.740225 163 over-trained
181 dense 4096 4096 1.0 1.834862 0.035995 -4.486367 163 over-trained
182 dense 4096 1024 4.0 1.921089 0.022146 -5.283070 75 over-trained
183 dense 4096 1024 4.0 1.840273 0.035097 -4.846709 62 over-trained
184 dense 4096 4096 1.0 1.868862 0.033259 -3.741389 159 over-trained
185 dense 4096 4096 1.0 1.756982 0.030109 -3.533943 159 over-trained
186 dense 4096 1024 4.0 2.006743 0.026763 -5.570579 56
187 dense 14336 4096 3.5 1.850240 0.026732 -2.548677 247 over-trained
188 dense 14336 4096 3.5 1.751108 0.021122 -2.694384 240 over-trained
189 dense 14336 4096 3.5 1.787283 0.028225 -2.462074 257 over-trained
190 dense 14336 4096 3.5 1.739146 0.022795 -2.599040 230 over-trained
191 dense 14336 4096 3.5 1.828319 0.029020 -2.396716 215 over-trained
192 dense 14336 4096 3.5 1.761196 0.029790 -2.352346 221 over-trained
193 dense 4096 1024 4.0 2.014984 0.044120 -5.931088 64
194 dense 4096 4096 1.0 1.739820 0.026572 -3.567883 142 over-trained
195 dense 4096 4096 1.0 1.971146 0.031445 -4.269282 39 over-trained
196 dense 4096 1024 4.0 1.821907 0.028822 -4.792173 59 over-trained
197 dense 4096 1024 4.0 1.897919 0.027914 -5.286584 64 over-trained
198 dense 4096 4096 1.0 1.855885 0.030338 -3.878162 53 over-trained
199 dense 4096 4096 1.0 1.750840 0.024849 -3.555933 149 over-trained
200 dense 14336 4096 3.5 1.810668 0.029999 -2.211924 202 over-trained
201 dense 14336 4096 3.5 1.757354 0.023843 -2.237877 200 over-trained
202 dense 14336 4096 3.5 1.736702 0.021952 -2.544528 217 over-trained
203 dense 4096 1024 4.0 1.950767 0.049588 -5.602196 64 over-trained
204 dense 14336 4096 3.5 1.713417 0.021265 -2.472632 181 over-trained
205 dense 14336 4096 3.5 1.781144 0.033185 -1.996601 173 over-trained
206 dense 14336 4096 3.5 1.755630 0.029540 -2.119852 186 over-trained
207 dense 4096 1024 4.0 1.905531 0.033908 -5.310633 59 over-trained
208 dense 4096 4096 1.0 1.717367 0.022643 -3.295417 123 over-trained
209 dense 4096 4096 1.0 1.790424 0.025244 -3.159784 114 over-trained
210 dense 4096 1024 4.0 1.857692 0.039074 -4.806877 67 over-trained
211 dense 4096 1024 4.0 1.870110 0.043617 -4.839119 58 over-trained
212 dense 4096 4096 1.0 1.744332 0.032964 -2.877448 121 over-trained
213 dense 4096 4096 1.0 1.689694 0.029231 -3.071418 114 over-trained
214 dense 14336 4096 3.5 1.738883 0.033170 -1.693919 175 over-trained
215 dense 14336 4096 3.5 1.692348 0.024472 -1.908499 162 over-trained
216 dense 14336 4096 3.5 1.623704 0.034947 -2.220283 138 over-trained
217 dense 4096 1024 4.0 1.945422 0.036302 -4.983935 82 over-trained
218 dense 4096 1024 4.0 1.791593 0.036860 -4.254341 66 over-trained
219 dense 14336 4096 3.5 1.680594 0.012986 -1.376911 123 over-trained
220 dense 14336 4096 3.5 1.639407 0.016393 -1.328007 127 over-trained
221 dense 4096 1024 4.0 1.788289 0.057287 -4.252953 50 over-trained
222 dense 4096 4096 1.0 1.632295 0.032135 -2.847892 95 over-trained
223 dense 4096 4096 1.0 1.717366 0.022736 -2.868483 99 over-trained
224 dense 14336 4096 3.5 1.555641 0.040652 -1.543000 131 over-trained