INDEX

Explanations

modal verbs and phrases indicating ability, possibility, necessity, and permission

oai_token-act-pair · gpt-3.5-turbo

negative contractions and expressions of inability or restriction

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

Chel

-0.68

 revisions

-0.64

burgh

-0.64

IVES

-0.62

iets

-0.62

REF

-0.61

Impro

-0.61

iyah

-0.60

 unchanged

-0.60

ridden

-0.59

POSITIVE LOGITS

 imagine

0.93

nikov

0.84

 guys

0.81

lihood

0.79

 yourselves

0.78

 yourself

0.76

ÅŁ

0.74

 wonder

0.70

think

0.69

 realise

0.68

Activations Density 0.367%

modal verbs and phrases indicating ability, possibility, necessity, and permission

negative contractions and expressions of inability or restriction

No Comments

No Known Activations