INDEX

Explanations

references to official documents and statements in various contexts

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

0:0.04

1:0.03

2:0.02

3:0.03

4:0.02

5:0.42

6:0.02

7:0.02

8:0.07

9:0.09

10:0.14

11:0.07

Negative Logits

idth

-1.86

icide

-1.83

icides

-1.75

bil

-1.73

phis

-1.71

 veto

-1.69

population

-1.66

iciary

-1.65

itans

-1.63

atu

-1.63

POSITIVE LOGITS

 dated

1.95

 titled

1.73

 excerpt

1.66

 Podesta

1.64

 recounted

1.63

 Letter

1.60

ographed

1.58

 excerpts

1.55

 outlines

1.55

 Blow

1.55

Activations Density 1.350%

references to official documents and statements in various contexts

No Comments

No Known Activations