INDEX

Explanations

phrases indicating the exclusion or dismissal of certain possibilities

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the denial or dismissal of possibilities

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

antry

-0.77

 conflic

-0.75

inki

-0.73

itialized

-0.72

elf

-0.72

iosyn

-0.71

awar

-0.70

omsky

-0.69

 resil

-0.69

feel

-0.67

POSITIVE LOGITS

posts

0.83

enance

0.71

nels

0.69

 anything

0.68

 pedestrians

0.67

stadt

0.64

 negatives

0.64

 dissenting

0.63

 Sussex

0.63

 Ernst

0.62

Activations Density 0.031%

phrases indicating the exclusion or dismissal of certain possibilities

phrases indicating the denial or dismissal of possibilities

No Comments

No Known Activations