INDEX

Explanations

terms related to negative judgment or condemnation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "foul" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

_>

-0.74

edia

-0.73

DCS

-0.71

Downloadha

-0.71

zl

-0.70

å§«

-0.69

 Airl

-0.68

itan

-0.68

etsk

-0.68

iku

-0.67

POSITIVE LOGITS

 foul

0.84

cery

0.81

 smelling

0.76

terness

0.70

 poisons

0.66

 stains

0.65

 misc

0.64

mire

0.64

fully

0.64

mson

0.64

Activations Density 0.006%

terms related to negative judgment or condemnation

instances of the word "foul" in various contexts

No Comments

No Known Activations