INDEX

Explanations

mentions of someone being in charge or having authority

oai_token-act-pair · gpt-3.5-turbo

the word "charge" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ĻĤ

-0.76

icz

-0.72

 livest

-0.69

pse

-0.69

 normalized

-0.68

psey

-0.68

wn

-0.66

pac

-0.66

akia

-0.65

 rook

-0.64

POSITIVE LOGITS

heet

0.97

AGES

0.92

Charges

0.90

abilities

0.87

Charge

0.86

 charges

0.85

 charge

0.83

charges

0.83

Charg

0.81

eters

0.79

Activations Density 0.025%

mentions of someone being in charge or having authority

the word "charge" in various contexts

No Comments

No Known Activations