|
<style type="text/css"> |
|
#T_a780d_row0_col2, #T_a780d_row0_col3, #T_a780d_row0_col6, #T_a780d_row0_col16, #T_a780d_row0_col17, #T_a780d_row1_col4, #T_a780d_row1_col5, #T_a780d_row1_col6, #T_a780d_row1_col7, #T_a780d_row1_col17, #T_a780d_row2_col4, #T_a780d_row2_col6, #T_a780d_row2_col7, #T_a780d_row2_col17, #T_a780d_row3_col6, #T_a780d_row3_col10, #T_a780d_row3_col11, #T_a780d_row3_col13, #T_a780d_row3_col17, #T_a780d_row4_col6, #T_a780d_row4_col9, #T_a780d_row4_col10, #T_a780d_row4_col11, #T_a780d_row4_col17, #T_a780d_row5_col6, #T_a780d_row5_col12, #T_a780d_row5_col17, #T_a780d_row6_col6, #T_a780d_row6_col8, #T_a780d_row6_col15, #T_a780d_row6_col17, #T_a780d_row7_col6, #T_a780d_row7_col17, #T_a780d_row8_col6, #T_a780d_row8_col17, #T_a780d_row9_col6, #T_a780d_row9_col17, #T_a780d_row10_col6, #T_a780d_row10_col14, #T_a780d_row10_col17, #T_a780d_row11_col6, #T_a780d_row11_col17 { |
|
background-color: #440154; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row0_col4, #T_a780d_row0_col7, #T_a780d_row0_col8, #T_a780d_row0_col9, #T_a780d_row0_col12, #T_a780d_row1_col13, #T_a780d_row1_col14, #T_a780d_row2_col15, #T_a780d_row9_col13, #T_a780d_row10_col13, #T_a780d_row11_col2, #T_a780d_row11_col3, #T_a780d_row11_col5, #T_a780d_row11_col10, #T_a780d_row11_col11, #T_a780d_row11_col13, #T_a780d_row11_col16 { |
|
background-color: #fde725; |
|
color: #000000; |
|
} |
|
#T_a780d_row0_col5 { |
|
background-color: #482576; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row0_col10 { |
|
background-color: #443a83; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row0_col11 { |
|
background-color: #414287; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row0_col13 { |
|
background-color: #fbe723; |
|
color: #000000; |
|
} |
|
#T_a780d_row0_col14 { |
|
background-color: #2e6f8e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row0_col15, #T_a780d_row6_col13 { |
|
background-color: #f4e61e; |
|
color: #000000; |
|
} |
|
#T_a780d_row1_col2, #T_a780d_row5_col16 { |
|
background-color: #482173; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row1_col3 { |
|
background-color: #48186a; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row1_col8 { |
|
background-color: #1e9d89; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row1_col9 { |
|
background-color: #d5e21a; |
|
color: #000000; |
|
} |
|
#T_a780d_row1_col10, #T_a780d_row2_col16, #T_a780d_row5_col10, #T_a780d_row10_col4 { |
|
background-color: #460a5d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row1_col11 { |
|
background-color: #470e61; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row1_col12 { |
|
background-color: #218f8d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row1_col15 { |
|
background-color: #4cc26c; |
|
color: #000000; |
|
} |
|
#T_a780d_row1_col16, #T_a780d_row3_col4, #T_a780d_row3_col7, #T_a780d_row4_col4, #T_a780d_row4_col7, #T_a780d_row5_col4, #T_a780d_row5_col7 { |
|
background-color: #440256; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col2, #T_a780d_row2_col3 { |
|
background-color: #433e85; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col5, #T_a780d_row4_col8 { |
|
background-color: #482677; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col8 { |
|
background-color: #2c738e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col9 { |
|
background-color: #44bf70; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col10 { |
|
background-color: #277e8e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col11 { |
|
background-color: #25838e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col12 { |
|
background-color: #31668e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row2_col13, #T_a780d_row4_col13 { |
|
background-color: #90d743; |
|
color: #000000; |
|
} |
|
#T_a780d_row2_col14 { |
|
background-color: #e7e419; |
|
color: #000000; |
|
} |
|
#T_a780d_row3_col2 { |
|
background-color: #38588c; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col3, #T_a780d_row9_col12 { |
|
background-color: #39568c; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col5 { |
|
background-color: #2b758e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col8 { |
|
background-color: #2f6c8e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col9, #T_a780d_row5_col9 { |
|
background-color: #33628d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col12 { |
|
background-color: #3f4889; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col14 { |
|
background-color: #d8e219; |
|
color: #000000; |
|
} |
|
#T_a780d_row3_col15 { |
|
background-color: #32658e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row3_col16, #T_a780d_row11_col4 { |
|
background-color: #471063; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row4_col2 { |
|
background-color: #2d708e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row4_col3 { |
|
background-color: #2e6e8e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row4_col5 { |
|
background-color: #21a685; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row4_col12, #T_a780d_row5_col15, #T_a780d_row6_col11, #T_a780d_row7_col15 { |
|
background-color: #471365; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row4_col14 { |
|
background-color: #95d840; |
|
color: #000000; |
|
} |
|
#T_a780d_row4_col15, #T_a780d_row9_col8 { |
|
background-color: #472a7a; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row4_col16, #T_a780d_row6_col10, #T_a780d_row7_col8, #T_a780d_row8_col15 { |
|
background-color: #481769; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row5_col2, #T_a780d_row11_col9 { |
|
background-color: #25858e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row5_col3 { |
|
background-color: #297b8e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row5_col5 { |
|
background-color: #24878e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row5_col8 { |
|
background-color: #481f70; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row5_col11, #T_a780d_row6_col12, #T_a780d_row9_col7 { |
|
background-color: #46075a; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row5_col13 { |
|
background-color: #8ed645; |
|
color: #000000; |
|
} |
|
#T_a780d_row5_col14 { |
|
background-color: #7fd34e; |
|
color: #000000; |
|
} |
|
#T_a780d_row6_col2 { |
|
background-color: #1e9b8a; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row6_col3, #T_a780d_row7_col3 { |
|
background-color: #1f968b; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row6_col4, #T_a780d_row6_col7, #T_a780d_row7_col4, #T_a780d_row7_col7, #T_a780d_row8_col7 { |
|
background-color: #450457; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row6_col5 { |
|
background-color: #46c06f; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row6_col9 { |
|
background-color: #481a6c; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row6_col14 { |
|
background-color: #56c667; |
|
color: #000000; |
|
} |
|
#T_a780d_row6_col16 { |
|
background-color: #472f7d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col2 { |
|
background-color: #2ab07f; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col5, #T_a780d_row11_col12 { |
|
background-color: #218e8d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col9, #T_a780d_row9_col10 { |
|
background-color: #34618d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col10 { |
|
background-color: #482878; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col11 { |
|
background-color: #482475; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col12 { |
|
background-color: #482071; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col13 { |
|
background-color: #f1e51d; |
|
color: #000000; |
|
} |
|
#T_a780d_row7_col14, #T_a780d_row8_col14 { |
|
background-color: #3aba76; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row7_col16, #T_a780d_row10_col8 { |
|
background-color: #424086; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col2 { |
|
background-color: #52c569; |
|
color: #000000; |
|
} |
|
#T_a780d_row8_col3 { |
|
background-color: #1fa188; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col4, #T_a780d_row8_col9, #T_a780d_row9_col4 { |
|
background-color: #450559; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col5 { |
|
background-color: #73d056; |
|
color: #000000; |
|
} |
|
#T_a780d_row8_col8 { |
|
background-color: #481c6e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col10 { |
|
background-color: #443b84; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col11 { |
|
background-color: #453781; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col12, #T_a780d_row9_col15 { |
|
background-color: #472c7a; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row8_col13 { |
|
background-color: #eae51a; |
|
color: #000000; |
|
} |
|
#T_a780d_row8_col16 { |
|
background-color: #38598c; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row9_col2 { |
|
background-color: #86d549; |
|
color: #000000; |
|
} |
|
#T_a780d_row9_col3 { |
|
background-color: #25ac82; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row9_col5 { |
|
background-color: #70cf57; |
|
color: #000000; |
|
} |
|
#T_a780d_row9_col9 { |
|
background-color: #433d84; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row9_col11, #T_a780d_row10_col15 { |
|
background-color: #365d8d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row9_col14 { |
|
background-color: #1fa287; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row9_col16 { |
|
background-color: #297a8e; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row10_col2 { |
|
background-color: #c2df23; |
|
color: #000000; |
|
} |
|
#T_a780d_row10_col3 { |
|
background-color: #65cb5e; |
|
color: #000000; |
|
} |
|
#T_a780d_row10_col5 { |
|
background-color: #93d741; |
|
color: #000000; |
|
} |
|
#T_a780d_row10_col7 { |
|
background-color: #46085c; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row10_col9 { |
|
background-color: #3e4c8a; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row10_col10 { |
|
background-color: #9dd93b; |
|
color: #000000; |
|
} |
|
#T_a780d_row10_col11 { |
|
background-color: #a8db34; |
|
color: #000000; |
|
} |
|
#T_a780d_row10_col12 { |
|
background-color: #54c568; |
|
color: #000000; |
|
} |
|
#T_a780d_row10_col16 { |
|
background-color: #3fbc73; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row11_col7 { |
|
background-color: #471164; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row11_col8 { |
|
background-color: #3b528b; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row11_col14 { |
|
background-color: #2db27d; |
|
color: #f1f1f1; |
|
} |
|
#T_a780d_row11_col15 { |
|
background-color: #23898e; |
|
color: #f1f1f1; |
|
} |
|
</style> |
|
<table id="T_a780d"> |
|
<thead> |
|
<tr> |
|
<th class="blank level0" > </th> |
|
<th id="T_a780d_level0_col0" class="col_heading level0 col0" >version</th> |
|
<th id="T_a780d_level0_col1" class="col_heading level0 col1" >d_sae</th> |
|
<th id="T_a780d_level0_col2" class="col_heading level0 col2" >layer</th> |
|
<th id="T_a780d_level0_col3" class="col_heading level0 col3" >kl_div_with_sae</th> |
|
<th id="T_a780d_level0_col4" class="col_heading level0 col4" >kl_div_with_ablation</th> |
|
<th id="T_a780d_level0_col5" class="col_heading level0 col5" >ce_loss_with_sae</th> |
|
<th id="T_a780d_level0_col6" class="col_heading level0 col6" >ce_loss_without_sae</th> |
|
<th id="T_a780d_level0_col7" class="col_heading level0 col7" >ce_loss_with_ablation</th> |
|
<th id="T_a780d_level0_col8" class="col_heading level0 col8" >kl_div_score</th> |
|
<th id="T_a780d_level0_col9" class="col_heading level0 col9" >ce_loss_score</th> |
|
<th id="T_a780d_level0_col10" class="col_heading level0 col10" >l2_norm_in</th> |
|
<th id="T_a780d_level0_col11" class="col_heading level0 col11" >l2_norm_out</th> |
|
<th id="T_a780d_level0_col12" class="col_heading level0 col12" >l2_ratio</th> |
|
<th id="T_a780d_level0_col13" class="col_heading level0 col13" >l0</th> |
|
<th id="T_a780d_level0_col14" class="col_heading level0 col14" >l1</th> |
|
<th id="T_a780d_level0_col15" class="col_heading level0 col15" >explained_variance</th> |
|
<th id="T_a780d_level0_col16" class="col_heading level0 col16" >mse</th> |
|
<th id="T_a780d_level0_col17" class="col_heading level0 col17" >total_tokens_evaluated</th> |
|
<th id="T_a780d_level0_col18" class="col_heading level0 col18" >filepath</th> |
|
</tr> |
|
</thead> |
|
<tbody> |
|
<tr> |
|
<th id="T_a780d_level0_row0" class="row_heading level0 row0" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_0/metrics.json</th> |
|
<td id="T_a780d_row0_col0" class="data row0 col0" >5</td> |
|
<td id="T_a780d_row0_col1" class="data row0 col1" >128</td> |
|
<td id="T_a780d_row0_col2" class="data row0 col2" >0</td> |
|
<td id="T_a780d_row0_col3" class="data row0 col3" >0.002883</td> |
|
<td id="T_a780d_row0_col4" class="data row0 col4" >3.094083</td> |
|
<td id="T_a780d_row0_col5" class="data row0 col5" >3.601894</td> |
|
<td id="T_a780d_row0_col6" class="data row0 col6" >3.599065</td> |
|
<td id="T_a780d_row0_col7" class="data row0 col7" >6.694649</td> |
|
<td id="T_a780d_row0_col8" class="data row0 col8" >0.999068</td> |
|
<td id="T_a780d_row0_col9" class="data row0 col9" >0.999086</td> |
|
<td id="T_a780d_row0_col10" class="data row0 col10" >29.933449</td> |
|
<td id="T_a780d_row0_col11" class="data row0 col11" >29.719198</td> |
|
<td id="T_a780d_row0_col12" class="data row0 col12" >0.993006</td> |
|
<td id="T_a780d_row0_col13" class="data row0 col13" >31.999350</td> |
|
<td id="T_a780d_row0_col14" class="data row0 col14" >61.198639</td> |
|
<td id="T_a780d_row0_col15" class="data row0 col15" >0.977519</td> |
|
<td id="T_a780d_row0_col16" class="data row0 col16" >13.276111</td> |
|
<td id="T_a780d_row0_col17" class="data row0 col17" >6144.000000</td> |
|
<td id="T_a780d_row0_col18" class="data row0 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_0/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row1" class="row_heading level0 row1" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_1/metrics.json</th> |
|
<td id="T_a780d_row1_col0" class="data row1 col0" >5</td> |
|
<td id="T_a780d_row1_col1" class="data row1 col1" >128</td> |
|
<td id="T_a780d_row1_col2" class="data row1 col2" >1</td> |
|
<td id="T_a780d_row1_col3" class="data row1 col3" >0.004242</td> |
|
<td id="T_a780d_row1_col4" class="data row1 col4" >0.051053</td> |
|
<td id="T_a780d_row1_col5" class="data row1 col5" >3.599821</td> |
|
<td id="T_a780d_row1_col6" class="data row1 col6" >3.599065</td> |
|
<td id="T_a780d_row1_col7" class="data row1 col7" >3.652537</td> |
|
<td id="T_a780d_row1_col8" class="data row1 col8" >0.916919</td> |
|
<td id="T_a780d_row1_col9" class="data row1 col9" >0.985857</td> |
|
<td id="T_a780d_row1_col10" class="data row1 col10" >18.973736</td> |
|
<td id="T_a780d_row1_col11" class="data row1 col11" >18.205364</td> |
|
<td id="T_a780d_row1_col12" class="data row1 col12" >0.932765</td> |
|
<td id="T_a780d_row1_col13" class="data row1 col13" >32.000000</td> |
|
<td id="T_a780d_row1_col14" class="data row1 col14" >84.060181</td> |
|
<td id="T_a780d_row1_col15" class="data row1 col15" >0.916917</td> |
|
<td id="T_a780d_row1_col16" class="data row1 col16" >17.682375</td> |
|
<td id="T_a780d_row1_col17" class="data row1 col17" >6144.000000</td> |
|
<td id="T_a780d_row1_col18" class="data row1 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_1/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row2" class="row_heading level0 row2" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_2/metrics.json</th> |
|
<td id="T_a780d_row2_col0" class="data row2 col0" >5</td> |
|
<td id="T_a780d_row2_col1" class="data row2 col1" >128</td> |
|
<td id="T_a780d_row2_col2" class="data row2 col2" >2</td> |
|
<td id="T_a780d_row2_col3" class="data row2 col3" >0.006752</td> |
|
<td id="T_a780d_row2_col4" class="data row2 col4" >0.058747</td> |
|
<td id="T_a780d_row2_col5" class="data row2 col5" >3.602034</td> |
|
<td id="T_a780d_row2_col6" class="data row2 col6" >3.599065</td> |
|
<td id="T_a780d_row2_col7" class="data row2 col7" >3.645913</td> |
|
<td id="T_a780d_row2_col8" class="data row2 col8" >0.885070</td> |
|
<td id="T_a780d_row2_col9" class="data row2 col9" >0.936614</td> |
|
<td id="T_a780d_row2_col10" class="data row2 col10" >49.106537</td> |
|
<td id="T_a780d_row2_col11" class="data row2 col11" >47.976685</td> |
|
<td id="T_a780d_row2_col12" class="data row2 col12" >0.912467</td> |
|
<td id="T_a780d_row2_col13" class="data row2 col13" >31.984375</td> |
|
<td id="T_a780d_row2_col14" class="data row2 col14" >82.676140</td> |
|
<td id="T_a780d_row2_col15" class="data row2 col15" >0.981090</td> |
|
<td id="T_a780d_row2_col16" class="data row2 col16" >28.213791</td> |
|
<td id="T_a780d_row2_col17" class="data row2 col17" >6144.000000</td> |
|
<td id="T_a780d_row2_col18" class="data row2 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_2/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row3" class="row_heading level0 row3" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_3/metrics.json</th> |
|
<td id="T_a780d_row3_col0" class="data row3 col0" >5</td> |
|
<td id="T_a780d_row3_col1" class="data row3 col1" >128</td> |
|
<td id="T_a780d_row3_col2" class="data row3 col2" >3</td> |
|
<td id="T_a780d_row3_col3" class="data row3 col3" >0.008532</td> |
|
<td id="T_a780d_row3_col4" class="data row3 col4" >0.070592</td> |
|
<td id="T_a780d_row3_col5" class="data row3 col5" >3.607723</td> |
|
<td id="T_a780d_row3_col6" class="data row3 col6" >3.599065</td> |
|
<td id="T_a780d_row3_col7" class="data row3 col7" >3.658678</td> |
|
<td id="T_a780d_row3_col8" class="data row3 col8" >0.879137</td> |
|
<td id="T_a780d_row3_col9" class="data row3 col9" >0.854756</td> |
|
<td id="T_a780d_row3_col10" class="data row3 col10" >16.987318</td> |
|
<td id="T_a780d_row3_col11" class="data row3 col11" >15.537837</td> |
|
<td id="T_a780d_row3_col12" class="data row3 col12" >0.899518</td> |
|
<td id="T_a780d_row3_col13" class="data row3 col13" >31.905111</td> |
|
<td id="T_a780d_row3_col14" class="data row3 col14" >81.886444</td> |
|
<td id="T_a780d_row3_col15" class="data row3 col15" >0.827494</td> |
|
<td id="T_a780d_row3_col16" class="data row3 col16" >38.814980</td> |
|
<td id="T_a780d_row3_col17" class="data row3 col17" >6144.000000</td> |
|
<td id="T_a780d_row3_col18" class="data row3 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_3/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row4" class="row_heading level0 row4" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_4/metrics.json</th> |
|
<td id="T_a780d_row4_col0" class="data row4 col0" >5</td> |
|
<td id="T_a780d_row4_col1" class="data row4 col1" >128</td> |
|
<td id="T_a780d_row4_col2" class="data row4 col2" >4</td> |
|
<td id="T_a780d_row4_col3" class="data row4 col3" >0.010456</td> |
|
<td id="T_a780d_row4_col4" class="data row4 col4" >0.063325</td> |
|
<td id="T_a780d_row4_col5" class="data row4 col5" >3.611892</td> |
|
<td id="T_a780d_row4_col6" class="data row4 col6" >3.599065</td> |
|
<td id="T_a780d_row4_col7" class="data row4 col7" >3.660080</td> |
|
<td id="T_a780d_row4_col8" class="data row4 col8" >0.834885</td> |
|
<td id="T_a780d_row4_col9" class="data row4 col9" >0.789762</td> |
|
<td id="T_a780d_row4_col10" class="data row4 col10" >17.251986</td> |
|
<td id="T_a780d_row4_col11" class="data row4 col11" >15.434065</td> |
|
<td id="T_a780d_row4_col12" class="data row4 col12" >0.879446</td> |
|
<td id="T_a780d_row4_col13" class="data row4 col13" >31.984375</td> |
|
<td id="T_a780d_row4_col14" class="data row4 col14" >78.371658</td> |
|
<td id="T_a780d_row4_col15" class="data row4 col15" >0.780778</td> |
|
<td id="T_a780d_row4_col16" class="data row4 col16" >50.818203</td> |
|
<td id="T_a780d_row4_col17" class="data row4 col17" >6144.000000</td> |
|
<td id="T_a780d_row4_col18" class="data row4 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_4/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row5" class="row_heading level0 row5" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_5/metrics.json</th> |
|
<td id="T_a780d_row5_col0" class="data row5 col0" >5</td> |
|
<td id="T_a780d_row5_col1" class="data row5 col1" >128</td> |
|
<td id="T_a780d_row5_col2" class="data row5 col2" >5</td> |
|
<td id="T_a780d_row5_col3" class="data row5 col3" >0.011618</td> |
|
<td id="T_a780d_row5_col4" class="data row5 col4" >0.068505</td> |
|
<td id="T_a780d_row5_col5" class="data row5 col5" >3.609259</td> |
|
<td id="T_a780d_row5_col6" class="data row5 col6" >3.599065</td> |
|
<td id="T_a780d_row5_col7" class="data row5 col7" >3.669386</td> |
|
<td id="T_a780d_row5_col8" class="data row5 col8" >0.830411</td> |
|
<td id="T_a780d_row5_col9" class="data row5 col9" >0.855026</td> |
|
<td id="T_a780d_row5_col10" class="data row5 col10" >18.888968</td> |
|
<td id="T_a780d_row5_col11" class="data row5 col11" >16.669851</td> |
|
<td id="T_a780d_row5_col12" class="data row5 col12" >0.873767</td> |
|
<td id="T_a780d_row5_col13" class="data row5 col13" >31.983725</td> |
|
<td id="T_a780d_row5_col14" class="data row5 col14" >77.139160</td> |
|
<td id="T_a780d_row5_col15" class="data row5 col15" >0.765372</td> |
|
<td id="T_a780d_row5_col16" class="data row5 col16" >71.502914</td> |
|
<td id="T_a780d_row5_col17" class="data row5 col17" >6144.000000</td> |
|
<td id="T_a780d_row5_col18" class="data row5 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_5/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row6" class="row_heading level0 row6" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_6/metrics.json</th> |
|
<td id="T_a780d_row6_col0" class="data row6 col0" >5</td> |
|
<td id="T_a780d_row6_col1" class="data row6 col1" >128</td> |
|
<td id="T_a780d_row6_col2" class="data row6 col2" >6</td> |
|
<td id="T_a780d_row6_col3" class="data row6 col3" >0.014023</td> |
|
<td id="T_a780d_row6_col4" class="data row6 col4" >0.075694</td> |
|
<td id="T_a780d_row6_col5" class="data row6 col5" >3.614241</td> |
|
<td id="T_a780d_row6_col6" class="data row6 col6" >3.599065</td> |
|
<td id="T_a780d_row6_col7" class="data row6 col7" >3.676516</td> |
|
<td id="T_a780d_row6_col8" class="data row6 col8" >0.814737</td> |
|
<td id="T_a780d_row6_col9" class="data row6 col9" >0.804058</td> |
|
<td id="T_a780d_row6_col10" class="data row6 col10" >21.466564</td> |
|
<td id="T_a780d_row6_col11" class="data row6 col11" >18.890602</td> |
|
<td id="T_a780d_row6_col12" class="data row6 col12" >0.875795</td> |
|
<td id="T_a780d_row6_col13" class="data row6 col13" >31.998373</td> |
|
<td id="T_a780d_row6_col14" class="data row6 col14" >74.617645</td> |
|
<td id="T_a780d_row6_col15" class="data row6 col15" >0.754052</td> |
|
<td id="T_a780d_row6_col16" class="data row6 col16" >97.221252</td> |
|
<td id="T_a780d_row6_col17" class="data row6 col17" >6144.000000</td> |
|
<td id="T_a780d_row6_col18" class="data row6 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_6/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row7" class="row_heading level0 row7" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_7/metrics.json</th> |
|
<td id="T_a780d_row7_col0" class="data row7 col0" >5</td> |
|
<td id="T_a780d_row7_col1" class="data row7 col1" >128</td> |
|
<td id="T_a780d_row7_col2" class="data row7 col2" >7</td> |
|
<td id="T_a780d_row7_col3" class="data row7 col3" >0.014005</td> |
|
<td id="T_a780d_row7_col4" class="data row7 col4" >0.080486</td> |
|
<td id="T_a780d_row7_col5" class="data row7 col5" >3.609832</td> |
|
<td id="T_a780d_row7_col6" class="data row7 col6" >3.599065</td> |
|
<td id="T_a780d_row7_col7" class="data row7 col7" >3.672712</td> |
|
<td id="T_a780d_row7_col8" class="data row7 col8" >0.825992</td> |
|
<td id="T_a780d_row7_col9" class="data row7 col9" >0.853803</td> |
|
<td id="T_a780d_row7_col10" class="data row7 col10" >25.444439</td> |
|
<td id="T_a780d_row7_col11" class="data row7 col11" >22.559195</td> |
|
<td id="T_a780d_row7_col12" class="data row7 col12" >0.884380</td> |
|
<td id="T_a780d_row7_col13" class="data row7 col13" >31.997885</td> |
|
<td id="T_a780d_row7_col14" class="data row7 col14" >72.551300</td> |
|
<td id="T_a780d_row7_col15" class="data row7 col15" >0.764806</td> |
|
<td id="T_a780d_row7_col16" class="data row7 col16" >130.780823</td> |
|
<td id="T_a780d_row7_col17" class="data row7 col17" >6144.000000</td> |
|
<td id="T_a780d_row7_col18" class="data row7 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_7/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row8" class="row_heading level0 row8" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_8/metrics.json</th> |
|
<td id="T_a780d_row8_col0" class="data row8 col0" >5</td> |
|
<td id="T_a780d_row8_col1" class="data row8 col1" >128</td> |
|
<td id="T_a780d_row8_col2" class="data row8 col2" >8</td> |
|
<td id="T_a780d_row8_col3" class="data row8 col3" >0.014951</td> |
|
<td id="T_a780d_row8_col4" class="data row8 col4" >0.087324</td> |
|
<td id="T_a780d_row8_col5" class="data row8 col5" >3.615898</td> |
|
<td id="T_a780d_row8_col6" class="data row8 col6" >3.599065</td> |
|
<td id="T_a780d_row8_col7" class="data row8 col7" >3.680337</td> |
|
<td id="T_a780d_row8_col8" class="data row8 col8" >0.828790</td> |
|
<td id="T_a780d_row8_col9" class="data row8 col9" >0.792874</td> |
|
<td id="T_a780d_row8_col10" class="data row8 col10" >30.250225</td> |
|
<td id="T_a780d_row8_col11" class="data row8 col11" >26.934555</td> |
|
<td id="T_a780d_row8_col12" class="data row8 col12" >0.888432</td> |
|
<td id="T_a780d_row8_col13" class="data row8 col13" >31.996908</td> |
|
<td id="T_a780d_row8_col14" class="data row8 col14" >72.559128</td> |
|
<td id="T_a780d_row8_col15" class="data row8 col15" >0.768030</td> |
|
<td id="T_a780d_row8_col16" class="data row8 col16" >183.634979</td> |
|
<td id="T_a780d_row8_col17" class="data row8 col17" >6144.000000</td> |
|
<td id="T_a780d_row8_col18" class="data row8 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_8/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row9" class="row_heading level0 row9" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_9/metrics.json</th> |
|
<td id="T_a780d_row9_col0" class="data row9 col0" >5</td> |
|
<td id="T_a780d_row9_col1" class="data row9 col1" >128</td> |
|
<td id="T_a780d_row9_col2" class="data row9 col2" >9</td> |
|
<td id="T_a780d_row9_col3" class="data row9 col3" >0.015919</td> |
|
<td id="T_a780d_row9_col4" class="data row9 col4" >0.097589</td> |
|
<td id="T_a780d_row9_col5" class="data row9 col5" >3.615844</td> |
|
<td id="T_a780d_row9_col6" class="data row9 col6" >3.599065</td> |
|
<td id="T_a780d_row9_col7" class="data row9 col7" >3.696245</td> |
|
<td id="T_a780d_row9_col8" class="data row9 col8" >0.836876</td> |
|
<td id="T_a780d_row9_col9" class="data row9 col9" >0.827333</td> |
|
<td id="T_a780d_row9_col10" class="data row9 col10" >40.192413</td> |
|
<td id="T_a780d_row9_col11" class="data row9 col11" >36.576889</td> |
|
<td id="T_a780d_row9_col12" class="data row9 col12" >0.905792</td> |
|
<td id="T_a780d_row9_col13" class="data row9 col13" >32.000000</td> |
|
<td id="T_a780d_row9_col14" class="data row9 col14" >68.976295</td> |
|
<td id="T_a780d_row9_col15" class="data row9 col15" >0.782096</td> |
|
<td id="T_a780d_row9_col16" class="data row9 col16" >267.216095</td> |
|
<td id="T_a780d_row9_col17" class="data row9 col17" >6144.000000</td> |
|
<td id="T_a780d_row9_col18" class="data row9 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_9/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row10" class="row_heading level0 row10" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_10/metrics.json</th> |
|
<td id="T_a780d_row10_col0" class="data row10 col0" >5</td> |
|
<td id="T_a780d_row10_col1" class="data row10 col1" >128</td> |
|
<td id="T_a780d_row10_col2" class="data row10 col2" >10</td> |
|
<td id="T_a780d_row10_col3" class="data row10 col3" >0.019077</td> |
|
<td id="T_a780d_row10_col4" class="data row10 col4" >0.126748</td> |
|
<td id="T_a780d_row10_col5" class="data row10 col5" >3.616906</td> |
|
<td id="T_a780d_row10_col6" class="data row10 col6" >3.599065</td> |
|
<td id="T_a780d_row10_col7" class="data row10 col7" >3.708984</td> |
|
<td id="T_a780d_row10_col8" class="data row10 col8" >0.849488</td> |
|
<td id="T_a780d_row10_col9" class="data row10 col9" >0.837687</td> |
|
<td id="T_a780d_row10_col10" class="data row10 col10" >81.756828</td> |
|
<td id="T_a780d_row10_col11" class="data row10 col11" >78.834885</td> |
|
<td id="T_a780d_row10_col12" class="data row10 col12" >0.960933</td> |
|
<td id="T_a780d_row10_col13" class="data row10 col13" >32.000000</td> |
|
<td id="T_a780d_row10_col14" class="data row10 col14" >48.323631</td> |
|
<td id="T_a780d_row10_col15" class="data row10 col15" >0.819787</td> |
|
<td id="T_a780d_row10_col16" class="data row10 col16" >443.108887</td> |
|
<td id="T_a780d_row10_col17" class="data row10 col17" >6144.000000</td> |
|
<td id="T_a780d_row10_col18" class="data row10 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_10/metrics.json</td> |
|
</tr> |
|
<tr> |
|
<th id="T_a780d_level0_row11" class="row_heading level0 row11" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_11/metrics.json</th> |
|
<td id="T_a780d_row11_col0" class="data row11 col0" >5</td> |
|
<td id="T_a780d_row11_col1" class="data row11 col1" >128</td> |
|
<td id="T_a780d_row11_col2" class="data row11 col2" >11</td> |
|
<td id="T_a780d_row11_col3" class="data row11 col3" >0.024072</td> |
|
<td id="T_a780d_row11_col4" class="data row11 col4" >0.173841</td> |
|
<td id="T_a780d_row11_col5" class="data row11 col5" >3.620230</td> |
|
<td id="T_a780d_row11_col6" class="data row11 col6" >3.599065</td> |
|
<td id="T_a780d_row11_col7" class="data row11 col7" >3.783318</td> |
|
<td id="T_a780d_row11_col8" class="data row11 col8" >0.861531</td> |
|
<td id="T_a780d_row11_col9" class="data row11 col9" >0.885130</td> |
|
<td id="T_a780d_row11_col10" class="data row11 col10" >92.906296</td> |
|
<td id="T_a780d_row11_col11" class="data row11 col11" >88.290031</td> |
|
<td id="T_a780d_row11_col12" class="data row11 col12" >0.932119</td> |
|
<td id="T_a780d_row11_col13" class="data row11 col13" >32.000000</td> |
|
<td id="T_a780d_row11_col14" class="data row11 col14" >71.255219</td> |
|
<td id="T_a780d_row11_col15" class="data row11 col15" >0.861334</td> |
|
<td id="T_a780d_row11_col16" class="data row11 col16" >635.459473</td> |
|
<td id="T_a780d_row11_col17" class="data row11 col17" >6144.000000</td> |
|
<td id="T_a780d_row11_col18" class="data row11 col18" >OAI_GPT2Small_v5_128k_resid_delta_mlp/v5_128k_layer_11/metrics.json</td> |
|
</tr> |
|
</tbody> |
|
</table> |
|
|