INDEX

Explanations

specific sections of text that reference titles, authors, or key details of literary works

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_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.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

0:0.27

1:0.13

2:0.04

3:0.06

4:0.04

5:0.07

6:0.04

7:0.02

8:0.04

9:0.08

10:0.09

11:0.06

Negative Logits

ETF

-1.85

 Tsukuyomi

-1.57

heter

-1.51

etsk

-1.49

Tam

-1.39

nce

-1.38

 stab

-1.38

uru

-1.36

 Tamil

-1.35

rave

-1.35

POSITIVE LOGITS

taboola

1.97

inspired

1.86

 Cosponsors

1.82

feat

1.81

metadata

1.66

by

1.64

��

1.61

 Featuring

1.60

icons

1.59

 Comed

1.58

Activations Density 0.002%

specific sections of text that reference titles, authors, or key details of literary works

No Comments

No Known Activations