GPT2-Small-OAI-v5-32k-attn-out-SAEs / benchmark_stats.html
jbloom's picture
Upload folder using huggingface_hub
6be0410 verified
raw
history blame
28.7 kB
<style type="text/css">
#T_b08a5_row0_col2, #T_b08a5_row0_col6, #T_b08a5_row0_col16, #T_b08a5_row0_col17, #T_b08a5_row1_col3, #T_b08a5_row1_col4, #T_b08a5_row1_col6, #T_b08a5_row1_col7, #T_b08a5_row1_col17, #T_b08a5_row2_col4, #T_b08a5_row2_col5, #T_b08a5_row2_col6, #T_b08a5_row2_col7, #T_b08a5_row2_col10, #T_b08a5_row2_col11, #T_b08a5_row2_col17, #T_b08a5_row3_col4, #T_b08a5_row3_col6, #T_b08a5_row3_col7, #T_b08a5_row3_col10, #T_b08a5_row3_col11, #T_b08a5_row3_col17, #T_b08a5_row4_col4, #T_b08a5_row4_col6, #T_b08a5_row4_col10, #T_b08a5_row4_col11, #T_b08a5_row4_col17, #T_b08a5_row5_col4, #T_b08a5_row5_col6, #T_b08a5_row5_col7, #T_b08a5_row5_col17, #T_b08a5_row6_col6, #T_b08a5_row6_col12, #T_b08a5_row6_col15, #T_b08a5_row6_col17, #T_b08a5_row7_col6, #T_b08a5_row7_col17, #T_b08a5_row8_col4, #T_b08a5_row8_col6, #T_b08a5_row8_col7, #T_b08a5_row8_col8, #T_b08a5_row8_col9, #T_b08a5_row8_col17, #T_b08a5_row9_col4, #T_b08a5_row9_col6, #T_b08a5_row9_col17, #T_b08a5_row10_col4, #T_b08a5_row10_col6, #T_b08a5_row10_col17, #T_b08a5_row11_col6, #T_b08a5_row11_col13, #T_b08a5_row11_col14, #T_b08a5_row11_col17 {
background-color: #440154;
color: #f1f1f1;
}
#T_b08a5_row0_col3, #T_b08a5_row7_col2 {
background-color: #2ab07f;
color: #f1f1f1;
}
#T_b08a5_row0_col4, #T_b08a5_row0_col7, #T_b08a5_row0_col8, #T_b08a5_row0_col9, #T_b08a5_row1_col13, #T_b08a5_row1_col14, #T_b08a5_row2_col13, #T_b08a5_row2_col14, #T_b08a5_row3_col13, #T_b08a5_row4_col13, #T_b08a5_row5_col13, #T_b08a5_row6_col13, #T_b08a5_row7_col5, #T_b08a5_row7_col13, #T_b08a5_row8_col3, #T_b08a5_row8_col13, #T_b08a5_row9_col13, #T_b08a5_row10_col13, #T_b08a5_row11_col2, #T_b08a5_row11_col10, #T_b08a5_row11_col11, #T_b08a5_row11_col12, #T_b08a5_row11_col15, #T_b08a5_row11_col16 {
background-color: #fde725;
color: #000000;
}
#T_b08a5_row0_col5 {
background-color: #37b878;
color: #f1f1f1;
}
#T_b08a5_row0_col10, #T_b08a5_row0_col11, #T_b08a5_row4_col15 {
background-color: #482071;
color: #f1f1f1;
}
#T_b08a5_row0_col12, #T_b08a5_row3_col14 {
background-color: #f1e51d;
color: #000000;
}
#T_b08a5_row0_col13 {
background-color: #f4e61e;
color: #000000;
}
#T_b08a5_row0_col14 {
background-color: #2c738e;
color: #f1f1f1;
}
#T_b08a5_row0_col15 {
background-color: #fbe723;
color: #000000;
}
#T_b08a5_row1_col2 {
background-color: #482173;
color: #f1f1f1;
}
#T_b08a5_row1_col5, #T_b08a5_row5_col16 {
background-color: #482374;
color: #f1f1f1;
}
#T_b08a5_row1_col8, #T_b08a5_row5_col3 {
background-color: #21a685;
color: #f1f1f1;
}
#T_b08a5_row1_col9, #T_b08a5_row5_col9 {
background-color: #20928c;
color: #f1f1f1;
}
#T_b08a5_row1_col10, #T_b08a5_row1_col11, #T_b08a5_row4_col7, #T_b08a5_row4_col12, #T_b08a5_row5_col10, #T_b08a5_row5_col11, #T_b08a5_row6_col4, #T_b08a5_row6_col7, #T_b08a5_row7_col4, #T_b08a5_row7_col7, #T_b08a5_row9_col7 {
background-color: #440256;
color: #f1f1f1;
}
#T_b08a5_row1_col12 {
background-color: #21918c;
color: #f1f1f1;
}
#T_b08a5_row1_col15, #T_b08a5_row4_col5 {
background-color: #1f988b;
color: #f1f1f1;
}
#T_b08a5_row1_col16, #T_b08a5_row2_col16, #T_b08a5_row6_col10, #T_b08a5_row6_col11, #T_b08a5_row10_col7 {
background-color: #450457;
color: #f1f1f1;
}
#T_b08a5_row2_col2 {
background-color: #433e85;
color: #f1f1f1;
}
#T_b08a5_row2_col3, #T_b08a5_row9_col15 {
background-color: #472d7b;
color: #f1f1f1;
}
#T_b08a5_row2_col8 {
background-color: #24aa83;
color: #f1f1f1;
}
#T_b08a5_row2_col9, #T_b08a5_row9_col9 {
background-color: #3dbc74;
color: #f1f1f1;
}
#T_b08a5_row2_col12 {
background-color: #27808e;
color: #f1f1f1;
}
#T_b08a5_row2_col15 {
background-color: #23888e;
color: #f1f1f1;
}
#T_b08a5_row3_col2, #T_b08a5_row5_col5 {
background-color: #38588c;
color: #f1f1f1;
}
#T_b08a5_row3_col3, #T_b08a5_row7_col8 {
background-color: #3a548c;
color: #f1f1f1;
}
#T_b08a5_row3_col5 {
background-color: #306a8e;
color: #f1f1f1;
}
#T_b08a5_row3_col8, #T_b08a5_row4_col9 {
background-color: #297a8e;
color: #f1f1f1;
}
#T_b08a5_row3_col9 {
background-color: #23898e;
color: #f1f1f1;
}
#T_b08a5_row3_col12 {
background-color: #3e4c8a;
color: #f1f1f1;
}
#T_b08a5_row3_col15 {
background-color: #365d8d;
color: #f1f1f1;
}
#T_b08a5_row3_col16 {
background-color: #470d60;
color: #f1f1f1;
}
#T_b08a5_row4_col2 {
background-color: #2d708e;
color: #f1f1f1;
}
#T_b08a5_row4_col3 {
background-color: #1f948c;
color: #f1f1f1;
}
#T_b08a5_row4_col8 {
background-color: #3b528b;
color: #f1f1f1;
}
#T_b08a5_row4_col14 {
background-color: #dae319;
color: #000000;
}
#T_b08a5_row4_col16 {
background-color: #481a6c;
color: #f1f1f1;
}
#T_b08a5_row5_col2 {
background-color: #25858e;
color: #f1f1f1;
}
#T_b08a5_row5_col8 {
background-color: #32648e;
color: #f1f1f1;
}
#T_b08a5_row5_col12 {
background-color: #481c6e;
color: #f1f1f1;
}
#T_b08a5_row5_col14 {
background-color: #cde11d;
color: #000000;
}
#T_b08a5_row5_col15 {
background-color: #472f7d;
color: #f1f1f1;
}
#T_b08a5_row6_col2 {
background-color: #1e9b8a;
color: #f1f1f1;
}
#T_b08a5_row6_col3 {
background-color: #a5db36;
color: #000000;
}
#T_b08a5_row6_col5 {
background-color: #81d34d;
color: #000000;
}
#T_b08a5_row6_col8, #T_b08a5_row7_col9 {
background-color: #423f85;
color: #f1f1f1;
}
#T_b08a5_row6_col9, #T_b08a5_row8_col16 {
background-color: #365c8d;
color: #f1f1f1;
}
#T_b08a5_row6_col14 {
background-color: #addc30;
color: #000000;
}
#T_b08a5_row6_col16, #T_b08a5_row8_col12, #T_b08a5_row9_col5 {
background-color: #453781;
color: #f1f1f1;
}
#T_b08a5_row7_col3 {
background-color: #84d44b;
color: #000000;
}
#T_b08a5_row7_col10, #T_b08a5_row7_col11 {
background-color: #46075a;
color: #f1f1f1;
}
#T_b08a5_row7_col12 {
background-color: #404688;
color: #f1f1f1;
}
#T_b08a5_row7_col14, #T_b08a5_row11_col8 {
background-color: #89d548;
color: #000000;
}
#T_b08a5_row7_col15 {
background-color: #46327e;
color: #f1f1f1;
}
#T_b08a5_row7_col16 {
background-color: #424186;
color: #f1f1f1;
}
#T_b08a5_row8_col2 {
background-color: #52c569;
color: #000000;
}
#T_b08a5_row8_col5 {
background-color: #b0dd2f;
color: #000000;
}
#T_b08a5_row8_col10 {
background-color: #460b5e;
color: #f1f1f1;
}
#T_b08a5_row8_col11, #T_b08a5_row8_col15 {
background-color: #460a5d;
color: #f1f1f1;
}
#T_b08a5_row8_col14 {
background-color: #8ed645;
color: #000000;
}
#T_b08a5_row9_col2 {
background-color: #86d549;
color: #000000;
}
#T_b08a5_row9_col3 {
background-color: #4ac16d;
color: #000000;
}
#T_b08a5_row9_col8 {
background-color: #433d84;
color: #f1f1f1;
}
#T_b08a5_row9_col10 {
background-color: #471164;
color: #f1f1f1;
}
#T_b08a5_row9_col11, #T_b08a5_row11_col4 {
background-color: #471063;
color: #f1f1f1;
}
#T_b08a5_row9_col12 {
background-color: #31688e;
color: #f1f1f1;
}
#T_b08a5_row9_col14 {
background-color: #58c765;
color: #000000;
}
#T_b08a5_row9_col16 {
background-color: #29798e;
color: #f1f1f1;
}
#T_b08a5_row10_col2 {
background-color: #c2df23;
color: #000000;
}
#T_b08a5_row10_col3 {
background-color: #1fa287;
color: #f1f1f1;
}
#T_b08a5_row10_col5 {
background-color: #287c8e;
color: #f1f1f1;
}
#T_b08a5_row10_col8 {
background-color: #46337f;
color: #f1f1f1;
}
#T_b08a5_row10_col9 {
background-color: #21a585;
color: #f1f1f1;
}
#T_b08a5_row10_col10 {
background-color: #481f70;
color: #f1f1f1;
}
#T_b08a5_row10_col11 {
background-color: #481d6f;
color: #f1f1f1;
}
#T_b08a5_row10_col12 {
background-color: #1fa088;
color: #f1f1f1;
}
#T_b08a5_row10_col14 {
background-color: #20a486;
color: #f1f1f1;
}
#T_b08a5_row10_col15 {
background-color: #34618d;
color: #f1f1f1;
}
#T_b08a5_row10_col16 {
background-color: #3bbb75;
color: #f1f1f1;
}
#T_b08a5_row11_col3 {
background-color: #20938c;
color: #f1f1f1;
}
#T_b08a5_row11_col5, #T_b08a5_row11_col7 {
background-color: #481467;
color: #f1f1f1;
}
#T_b08a5_row11_col9 {
background-color: #d5e21a;
color: #000000;
}
</style>
<table id="T_b08a5">
<thead>
<tr>
<th class="blank level0" >&nbsp;</th>
<th id="T_b08a5_level0_col0" class="col_heading level0 col0" >version</th>
<th id="T_b08a5_level0_col1" class="col_heading level0 col1" >d_sae</th>
<th id="T_b08a5_level0_col2" class="col_heading level0 col2" >layer</th>
<th id="T_b08a5_level0_col3" class="col_heading level0 col3" >kl_div_with_sae</th>
<th id="T_b08a5_level0_col4" class="col_heading level0 col4" >kl_div_with_ablation</th>
<th id="T_b08a5_level0_col5" class="col_heading level0 col5" >ce_loss_with_sae</th>
<th id="T_b08a5_level0_col6" class="col_heading level0 col6" >ce_loss_without_sae</th>
<th id="T_b08a5_level0_col7" class="col_heading level0 col7" >ce_loss_with_ablation</th>
<th id="T_b08a5_level0_col8" class="col_heading level0 col8" >kl_div_score</th>
<th id="T_b08a5_level0_col9" class="col_heading level0 col9" >ce_loss_score</th>
<th id="T_b08a5_level0_col10" class="col_heading level0 col10" >l2_norm_in</th>
<th id="T_b08a5_level0_col11" class="col_heading level0 col11" >l2_norm_out</th>
<th id="T_b08a5_level0_col12" class="col_heading level0 col12" >l2_ratio</th>
<th id="T_b08a5_level0_col13" class="col_heading level0 col13" >l0</th>
<th id="T_b08a5_level0_col14" class="col_heading level0 col14" >l1</th>
<th id="T_b08a5_level0_col15" class="col_heading level0 col15" >explained_variance</th>
<th id="T_b08a5_level0_col16" class="col_heading level0 col16" >mse</th>
<th id="T_b08a5_level0_col17" class="col_heading level0 col17" >total_tokens_evaluated</th>
<th id="T_b08a5_level0_col18" class="col_heading level0 col18" >filepath</th>
</tr>
</thead>
<tbody>
<tr>
<th id="T_b08a5_level0_row0" class="row_heading level0 row0" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_0/metrics.json</th>
<td id="T_b08a5_row0_col0" class="data row0 col0" >5</td>
<td id="T_b08a5_row0_col1" class="data row0 col1" >32</td>
<td id="T_b08a5_row0_col2" class="data row0 col2" >0</td>
<td id="T_b08a5_row0_col3" class="data row0 col3" >0.004215</td>
<td id="T_b08a5_row0_col4" class="data row0 col4" >2.121528</td>
<td id="T_b08a5_row0_col5" class="data row0 col5" >3.603763</td>
<td id="T_b08a5_row0_col6" class="data row0 col6" >3.599065</td>
<td id="T_b08a5_row0_col7" class="data row0 col7" >5.748601</td>
<td id="T_b08a5_row0_col8" class="data row0 col8" >0.998013</td>
<td id="T_b08a5_row0_col9" class="data row0 col9" >0.997814</td>
<td id="T_b08a5_row0_col10" class="data row0 col10" >32.013138</td>
<td id="T_b08a5_row0_col11" class="data row0 col11" >31.891546</td>
<td id="T_b08a5_row0_col12" class="data row0 col12" >0.996240</td>
<td id="T_b08a5_row0_col13" class="data row0 col13" >31.993979</td>
<td id="T_b08a5_row0_col14" class="data row0 col14" >42.146969</td>
<td id="T_b08a5_row0_col15" class="data row0 col15" >0.966712</td>
<td id="T_b08a5_row0_col16" class="data row0 col16" >8.074387</td>
<td id="T_b08a5_row0_col17" class="data row0 col17" >6144.000000</td>
<td id="T_b08a5_row0_col18" class="data row0 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_0/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row1" class="row_heading level0 row1" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_1/metrics.json</th>
<td id="T_b08a5_row1_col0" class="data row1 col0" >5</td>
<td id="T_b08a5_row1_col1" class="data row1 col1" >32</td>
<td id="T_b08a5_row1_col2" class="data row1 col2" >1</td>
<td id="T_b08a5_row1_col3" class="data row1 col3" >0.001881</td>
<td id="T_b08a5_row1_col4" class="data row1 col4" >0.024066</td>
<td id="T_b08a5_row1_col5" class="data row1 col5" >3.601315</td>
<td id="T_b08a5_row1_col6" class="data row1 col6" >3.599065</td>
<td id="T_b08a5_row1_col7" class="data row1 col7" >3.620633</td>
<td id="T_b08a5_row1_col8" class="data row1 col8" >0.921821</td>
<td id="T_b08a5_row1_col9" class="data row1 col9" >0.895681</td>
<td id="T_b08a5_row1_col10" class="data row1 col10" >9.714648</td>
<td id="T_b08a5_row1_col11" class="data row1 col11" >9.157854</td>
<td id="T_b08a5_row1_col12" class="data row1 col12" >0.937912</td>
<td id="T_b08a5_row1_col13" class="data row1 col13" >32.000000</td>
<td id="T_b08a5_row1_col14" class="data row1 col14" >82.860558</td>
<td id="T_b08a5_row1_col15" class="data row1 col15" >0.868087</td>
<td id="T_b08a5_row1_col16" class="data row1 col16" >9.541979</td>
<td id="T_b08a5_row1_col17" class="data row1 col17" >6144.000000</td>
<td id="T_b08a5_row1_col18" class="data row1 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_1/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row2" class="row_heading level0 row2" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_2/metrics.json</th>
<td id="T_b08a5_row2_col0" class="data row2 col0" >5</td>
<td id="T_b08a5_row2_col1" class="data row2 col1" >32</td>
<td id="T_b08a5_row2_col2" class="data row2 col2" >2</td>
<td id="T_b08a5_row2_col3" class="data row2 col3" >0.002342</td>
<td id="T_b08a5_row2_col4" class="data row2 col4" >0.031005</td>
<td id="T_b08a5_row2_col5" class="data row2 col5" >3.600916</td>
<td id="T_b08a5_row2_col6" class="data row2 col6" >3.599065</td>
<td id="T_b08a5_row2_col7" class="data row2 col7" >3.626660</td>
<td id="T_b08a5_row2_col8" class="data row2 col8" >0.924478</td>
<td id="T_b08a5_row2_col9" class="data row2 col9" >0.932912</td>
<td id="T_b08a5_row2_col10" class="data row2 col10" >8.641823</td>
<td id="T_b08a5_row2_col11" class="data row2 col11" >8.045539</td>
<td id="T_b08a5_row2_col12" class="data row2 col12" >0.929619</td>
<td id="T_b08a5_row2_col13" class="data row2 col13" >32.000000</td>
<td id="T_b08a5_row2_col14" class="data row2 col14" >82.863617</td>
<td id="T_b08a5_row2_col15" class="data row2 col15" >0.853233</td>
<td id="T_b08a5_row2_col16" class="data row2 col16" >9.740603</td>
<td id="T_b08a5_row2_col17" class="data row2 col17" >6144.000000</td>
<td id="T_b08a5_row2_col18" class="data row2 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_2/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row3" class="row_heading level0 row3" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_3/metrics.json</th>
<td id="T_b08a5_row3_col0" class="data row3 col0" >5</td>
<td id="T_b08a5_row3_col1" class="data row3 col1" >32</td>
<td id="T_b08a5_row3_col2" class="data row3 col2" >3</td>
<td id="T_b08a5_row3_col3" class="data row3 col3" >0.002842</td>
<td id="T_b08a5_row3_col4" class="data row3 col4" >0.025134</td>
<td id="T_b08a5_row3_col5" class="data row3 col5" >3.602360</td>
<td id="T_b08a5_row3_col6" class="data row3 col6" >3.599065</td>
<td id="T_b08a5_row3_col7" class="data row3 col7" >3.628661</td>
<td id="T_b08a5_row3_col8" class="data row3 col8" >0.886914</td>
<td id="T_b08a5_row3_col9" class="data row3 col9" >0.888638</td>
<td id="T_b08a5_row3_col10" class="data row3 col10" >8.571012</td>
<td id="T_b08a5_row3_col11" class="data row3 col11" >7.753783</td>
<td id="T_b08a5_row3_col12" class="data row3 col12" >0.904760</td>
<td id="T_b08a5_row3_col13" class="data row3 col13" >32.000000</td>
<td id="T_b08a5_row3_col14" class="data row3 col14" >81.563385</td>
<td id="T_b08a5_row3_col15" class="data row3 col15" >0.815689</td>
<td id="T_b08a5_row3_col16" class="data row3 col16" >13.545696</td>
<td id="T_b08a5_row3_col17" class="data row3 col17" >6144.000000</td>
<td id="T_b08a5_row3_col18" class="data row3 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_3/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row4" class="row_heading level0 row4" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_4/metrics.json</th>
<td id="T_b08a5_row4_col0" class="data row4 col0" >5</td>
<td id="T_b08a5_row4_col1" class="data row4 col1" >32</td>
<td id="T_b08a5_row4_col2" class="data row4 col2" >4</td>
<td id="T_b08a5_row4_col3" class="data row4 col3" >0.003790</td>
<td id="T_b08a5_row4_col4" class="data row4 col4" >0.026723</td>
<td id="T_b08a5_row4_col5" class="data row4 col5" >3.603180</td>
<td id="T_b08a5_row4_col6" class="data row4 col6" >3.599065</td>
<td id="T_b08a5_row4_col7" class="data row4 col7" >3.632133</td>
<td id="T_b08a5_row4_col8" class="data row4 col8" >0.858169</td>
<td id="T_b08a5_row4_col9" class="data row4 col9" >0.875538</td>
<td id="T_b08a5_row4_col10" class="data row4 col10" >9.123016</td>
<td id="T_b08a5_row4_col11" class="data row4 col11" >7.993571</td>
<td id="T_b08a5_row4_col12" class="data row4 col12" >0.877799</td>
<td id="T_b08a5_row4_col13" class="data row4 col13" >32.000000</td>
<td id="T_b08a5_row4_col14" class="data row4 col14" >79.127533</td>
<td id="T_b08a5_row4_col15" class="data row4 col15" >0.772355</td>
<td id="T_b08a5_row4_col16" class="data row4 col16" >19.599684</td>
<td id="T_b08a5_row4_col17" class="data row4 col17" >6144.000000</td>
<td id="T_b08a5_row4_col18" class="data row4 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_4/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row5" class="row_heading level0 row5" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_5/metrics.json</th>
<td id="T_b08a5_row5_col0" class="data row5 col0" >5</td>
<td id="T_b08a5_row5_col1" class="data row5 col1" >32</td>
<td id="T_b08a5_row5_col2" class="data row5 col2" >5</td>
<td id="T_b08a5_row5_col3" class="data row5 col3" >0.004055</td>
<td id="T_b08a5_row5_col4" class="data row5 col4" >0.031378</td>
<td id="T_b08a5_row5_col5" class="data row5 col5" >3.602063</td>
<td id="T_b08a5_row5_col6" class="data row5 col6" >3.599065</td>
<td id="T_b08a5_row5_col7" class="data row5 col7" >3.627760</td>
<td id="T_b08a5_row5_col8" class="data row5 col8" >0.870766</td>
<td id="T_b08a5_row5_col9" class="data row5 col9" >0.895510</td>
<td id="T_b08a5_row5_col10" class="data row5 col10" >10.034396</td>
<td id="T_b08a5_row5_col11" class="data row5 col11" >8.880257</td>
<td id="T_b08a5_row5_col12" class="data row5 col12" >0.886243</td>
<td id="T_b08a5_row5_col13" class="data row5 col13" >32.000000</td>
<td id="T_b08a5_row5_col14" class="data row5 col14" >77.734062</td>
<td id="T_b08a5_row5_col15" class="data row5 col15" >0.782550</td>
<td id="T_b08a5_row5_col16" class="data row5 col16" >24.739639</td>
<td id="T_b08a5_row5_col17" class="data row5 col17" >6144.000000</td>
<td id="T_b08a5_row5_col18" class="data row5 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_5/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row6" class="row_heading level0 row6" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_6/metrics.json</th>
<td id="T_b08a5_row6_col0" class="data row6 col0" >5</td>
<td id="T_b08a5_row6_col1" class="data row6 col1" >32</td>
<td id="T_b08a5_row6_col2" class="data row6 col2" >6</td>
<td id="T_b08a5_row6_col3" class="data row6 col3" >0.005056</td>
<td id="T_b08a5_row6_col4" class="data row6 col4" >0.032789</td>
<td id="T_b08a5_row6_col5" class="data row6 col5" >3.604352</td>
<td id="T_b08a5_row6_col6" class="data row6 col6" >3.599065</td>
<td id="T_b08a5_row6_col7" class="data row6 col7" >3.634286</td>
<td id="T_b08a5_row6_col8" class="data row6 col8" >0.845789</td>
<td id="T_b08a5_row6_col9" class="data row6 col9" >0.849889</td>
<td id="T_b08a5_row6_col10" class="data row6 col10" >11.678066</td>
<td id="T_b08a5_row6_col11" class="data row6 col11" >10.210135</td>
<td id="T_b08a5_row6_col12" class="data row6 col12" >0.877007</td>
<td id="T_b08a5_row6_col13" class="data row6 col13" >32.000000</td>
<td id="T_b08a5_row6_col14" class="data row6 col14" >74.857086</td>
<td id="T_b08a5_row6_col15" class="data row6 col15" >0.753413</td>
<td id="T_b08a5_row6_col16" class="data row6 col16" >35.039909</td>
<td id="T_b08a5_row6_col17" class="data row6 col17" >6144.000000</td>
<td id="T_b08a5_row6_col18" class="data row6 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_6/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row7" class="row_heading level0 row7" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_7/metrics.json</th>
<td id="T_b08a5_row7_col0" class="data row7 col0" >5</td>
<td id="T_b08a5_row7_col1" class="data row7 col1" >32</td>
<td id="T_b08a5_row7_col2" class="data row7 col2" >7</td>
<td id="T_b08a5_row7_col3" class="data row7 col3" >0.004876</td>
<td id="T_b08a5_row7_col4" class="data row7 col4" >0.034661</td>
<td id="T_b08a5_row7_col5" class="data row7 col5" >3.605160</td>
<td id="T_b08a5_row7_col6" class="data row7 col6" >3.599065</td>
<td id="T_b08a5_row7_col7" class="data row7 col7" >3.634834</td>
<td id="T_b08a5_row7_col8" class="data row7 col8" >0.859336</td>
<td id="T_b08a5_row7_col9" class="data row7 col9" >0.829584</td>
<td id="T_b08a5_row7_col10" class="data row7 col10" >13.650208</td>
<td id="T_b08a5_row7_col11" class="data row7 col11" >12.291288</td>
<td id="T_b08a5_row7_col12" class="data row7 col12" >0.902289</td>
<td id="T_b08a5_row7_col13" class="data row7 col13" >32.000000</td>
<td id="T_b08a5_row7_col14" class="data row7 col14" >71.106369</td>
<td id="T_b08a5_row7_col15" class="data row7 col15" >0.783828</td>
<td id="T_b08a5_row7_col16" class="data row7 col16" >41.602531</td>
<td id="T_b08a5_row7_col17" class="data row7 col17" >6144.000000</td>
<td id="T_b08a5_row7_col18" class="data row7 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_7/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row8" class="row_heading level0 row8" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_8/metrics.json</th>
<td id="T_b08a5_row8_col0" class="data row8 col0" >5</td>
<td id="T_b08a5_row8_col1" class="data row8 col1" >32</td>
<td id="T_b08a5_row8_col2" class="data row8 col2" >8</td>
<td id="T_b08a5_row8_col3" class="data row8 col3" >0.005558</td>
<td id="T_b08a5_row8_col4" class="data row8 col4" >0.029382</td>
<td id="T_b08a5_row8_col5" class="data row8 col5" >3.604657</td>
<td id="T_b08a5_row8_col6" class="data row8 col6" >3.599065</td>
<td id="T_b08a5_row8_col7" class="data row8 col7" >3.625802</td>
<td id="T_b08a5_row8_col8" class="data row8 col8" >0.810828</td>
<td id="T_b08a5_row8_col9" class="data row8 col9" >0.790831</td>
<td id="T_b08a5_row8_col10" class="data row8 col10" >16.137949</td>
<td id="T_b08a5_row8_col11" class="data row8 col11" >14.443828</td>
<td id="T_b08a5_row8_col12" class="data row8 col12" >0.896318</td>
<td id="T_b08a5_row8_col13" class="data row8 col13" >32.000000</td>
<td id="T_b08a5_row8_col14" class="data row8 col14" >71.654037</td>
<td id="T_b08a5_row8_col15" class="data row8 col15" >0.759009</td>
<td id="T_b08a5_row8_col16" class="data row8 col16" >57.211456</td>
<td id="T_b08a5_row8_col17" class="data row8 col17" >6144.000000</td>
<td id="T_b08a5_row8_col18" class="data row8 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_8/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row9" class="row_heading level0 row9" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_9/metrics.json</th>
<td id="T_b08a5_row9_col0" class="data row9 col0" >5</td>
<td id="T_b08a5_row9_col1" class="data row9 col1" >32</td>
<td id="T_b08a5_row9_col2" class="data row9 col2" >9</td>
<td id="T_b08a5_row9_col3" class="data row9 col3" >0.004499</td>
<td id="T_b08a5_row9_col4" class="data row9 col4" >0.028918</td>
<td id="T_b08a5_row9_col5" class="data row9 col5" >3.601583</td>
<td id="T_b08a5_row9_col6" class="data row9 col6" >3.599065</td>
<td id="T_b08a5_row9_col7" class="data row9 col7" >3.636501</td>
<td id="T_b08a5_row9_col8" class="data row9 col8" >0.844428</td>
<td id="T_b08a5_row9_col9" class="data row9 col9" >0.932728</td>
<td id="T_b08a5_row9_col10" class="data row9 col10" >20.912498</td>
<td id="T_b08a5_row9_col11" class="data row9 col11" >19.139347</td>
<td id="T_b08a5_row9_col12" class="data row9 col12" >0.917789</td>
<td id="T_b08a5_row9_col13" class="data row9 col13" >32.000000</td>
<td id="T_b08a5_row9_col14" class="data row9 col14" >65.829063</td>
<td id="T_b08a5_row9_col15" class="data row9 col15" >0.780790</td>
<td id="T_b08a5_row9_col16" class="data row9 col16" >77.843826</td>
<td id="T_b08a5_row9_col17" class="data row9 col17" >6144.000000</td>
<td id="T_b08a5_row9_col18" class="data row9 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_9/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row10" class="row_heading level0 row10" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_10/metrics.json</th>
<td id="T_b08a5_row10_col0" class="data row10 col0" >5</td>
<td id="T_b08a5_row10_col1" class="data row10 col1" >32</td>
<td id="T_b08a5_row10_col2" class="data row10 col2" >10</td>
<td id="T_b08a5_row10_col3" class="data row10 col3" >0.003999</td>
<td id="T_b08a5_row10_col4" class="data row10 col4" >0.024752</td>
<td id="T_b08a5_row10_col5" class="data row10 col5" >3.602678</td>
<td id="T_b08a5_row10_col6" class="data row10 col6" >3.599065</td>
<td id="T_b08a5_row10_col7" class="data row10 col7" >3.640488</td>
<td id="T_b08a5_row10_col8" class="data row10 col8" >0.838441</td>
<td id="T_b08a5_row10_col9" class="data row10 col9" >0.912773</td>
<td id="T_b08a5_row10_col10" class="data row10 col10" >31.821377</td>
<td id="T_b08a5_row10_col11" class="data row10 col11" >30.121130</td>
<td id="T_b08a5_row10_col12" class="data row10 col12" >0.945757</td>
<td id="T_b08a5_row10_col13" class="data row10 col13" >32.000000</td>
<td id="T_b08a5_row10_col14" class="data row10 col14" >55.563881</td>
<td id="T_b08a5_row10_col15" class="data row10 col15" >0.819294</td>
<td id="T_b08a5_row10_col16" class="data row10 col16" >125.342606</td>
<td id="T_b08a5_row10_col17" class="data row10 col17" >6144.000000</td>
<td id="T_b08a5_row10_col18" class="data row10 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_10/metrics.json</td>
</tr>
<tr>
<th id="T_b08a5_level0_row11" class="row_heading level0 row11" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_11/metrics.json</th>
<td id="T_b08a5_row11_col0" class="data row11 col0" >5</td>
<td id="T_b08a5_row11_col1" class="data row11 col1" >32</td>
<td id="T_b08a5_row11_col2" class="data row11 col2" >11</td>
<td id="T_b08a5_row11_col3" class="data row11 col3" >0.003772</td>
<td id="T_b08a5_row11_col4" class="data row11 col4" >0.106875</td>
<td id="T_b08a5_row11_col5" class="data row11 col5" >3.601134</td>
<td id="T_b08a5_row11_col6" class="data row11 col6" >3.599065</td>
<td id="T_b08a5_row11_col7" class="data row11 col7" >3.730870</td>
<td id="T_b08a5_row11_col8" class="data row11 col8" >0.964708</td>
<td id="T_b08a5_row11_col9" class="data row11 col9" >0.984299</td>
<td id="T_b08a5_row11_col10" class="data row11 col10" >280.864441</td>
<td id="T_b08a5_row11_col11" class="data row11 col11" >280.543213</td>
<td id="T_b08a5_row11_col12" class="data row11 col12" >0.998668</td>
<td id="T_b08a5_row11_col13" class="data row11 col13" >31.687500</td>
<td id="T_b08a5_row11_col14" class="data row11 col14" >17.145309</td>
<td id="T_b08a5_row11_col15" class="data row11 col15" >0.967849</td>
<td id="T_b08a5_row11_col16" class="data row11 col16" >180.271027</td>
<td id="T_b08a5_row11_col17" class="data row11 col17" >6144.000000</td>
<td id="T_b08a5_row11_col18" class="data row11 col18" >OAI_GPT2Small_v5_32k_resid_delta_attn/v5_32k_layer_11/metrics.json</td>
</tr>
</tbody>
</table>