INDEX

Explanations

phrases related to regular operations or procedures

oai_token-act-pair · gpt-3.5-turbo

standard references to established norms or procedures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

psey

-0.78

ificantly

-0.77

arching

-0.75

untarily

-0.71

antically

-0.70

anamo

-0.65

amia

-0.65

ependent

-0.65

 Canaver

-0.64

olulu

-0.63

POSITIVE LOGITS

 fare

0.80

 decency

0.77

 tropes

0.74

 greeting

0.72

 procedure

0.71

 guiActiveUnfocused

0.67

ItemImage

0.67

ensical

0.67

iquette

0.66

 alike

0.65

Activations Density 0.439%

phrases related to regular operations or procedures

standard references to established norms or procedures

No Comments

No Known Activations