INDEX

Explanations

proper nouns ending in 'ker' with varying degrees of specificity

oai_token-act-pair · gpt-3.5-turbo

occurrences of the suffix "ker" in various forms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Cind

-0.75

UME

-0.73

erest

-0.71

eree

-0.64

 haste

-0.64

ancies

-0.61

IRD

-0.60

UE

-0.59

ental

-0.59

ERROR

-0.59

POSITIVE LOGITS

chief

1.07

ker

1.02

lund

0.93

lar

0.87

lein

0.86

icker

0.84

geon

0.82

pedia

0.82

rigan

0.81

king

0.80

Activations Density 0.015%

proper nouns ending in 'ker' with varying degrees of specificity

occurrences of the suffix "ker" in various forms

No Comments

No Known Activations