INDEX

Explanations

phrases related to outlining complex ideas and details

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.12

2:0.11

3:0.04

4:0.03

5:0.06

6:0.10

7:0.09

8:0.08

9:0.08

10:0.07

11:0.15

Negative Logits

 honoured

-1.33

 honored

-1.21

 preserved

-1.18

 complied

-1.16

handled

-1.16

 compounded

-1.12

 overturned

-1.12

 condol

-1.10

 retained

-1.09

 tolerated

-1.08

POSITIVE LOGITS

¯

1.37

 Strongh

1.33

 conclusions

1.25

 Spells

1.20

elo

1.19

ologies

1.14

 Conclusion

1.13

 Judgment

1.13

ーティ

1.12

 ourselves

1.10

Activations Density 0.010%

phrases related to outlining complex ideas and details

No Comments

No Known Activations