INDEX

Explanations

locations or proximity related terms

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "near."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ÑĮ

-0.79

onew

-0.79

displayText

-0.78

rosse

-0.74

é»Ĵ

-0.71

IAL

-0.71

 externalToEVAOnly

-0.71

éĹĺ

-0.70

ILLE

-0.68

istration

-0.68

POSITIVE LOGITS

 Near

1.05

entimes

0.85

sighted

0.84

vine

0.79

Near

0.74

uve

0.73

qus

0.71

oken

0.69

 misses

0.69

cuts

0.69

Activations Density 0.009%

locations or proximity related terms

mentions of the term "near."

No Comments

No Known Activations