INDEX

Explanations

phrases related to political entities such as "bloc" and "borough"

oai_token-act-pair · gpt-3.5-turbo

references to political or organizational groups

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

sbm

-0.74

HOME

-0.68

ãĥī

-0.67

Condition

-0.67

Nusra

-0.65

Course

-0.64

Adv

-0.64

 Qiao

-0.63

Journal

-0.63

mony

-0.61

POSITIVE LOGITS

 membership

0.79

0.78

busters

0.75

 bloc

0.74

 affili

0.73

enary

0.70

nuts

0.69

 converter

0.69

 elector

0.68

aby

0.68

Activations Density 0.014%

phrases related to political entities such as "bloc" and "borough"

references to political or organizational groups

No Comments

No Known Activations