INDEX

Explanations

phrases related to retracting or withdrawing statements or actions

oai_token-act-pair · gpt-3.5-turbo

terms related to the retraction or removal of statements, articles, or support

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

idth

-0.76

ixture

-0.74

iour

-0.73

ixtures

-0.71

ancial

-0.71

aez

-0.68

irit

-0.67

icult

-0.67

achine

-0.64

inav

-0.63

POSITIVE LOGITS

 abruptly

0.90

 outright

0.87

 unanimously

0.82

 laughing

0.75

dated

0.73

 recommending

0.72

 disbelief

0.71

 prematurely

0.71

 unres

0.71

 aback

0.70

Activations Density 0.145%

phrases related to retracting or withdrawing statements or actions

terms related to the retraction or removal of statements, articles, or support

No Comments

No Known Activations