INDEX

Explanations

adjectives with negative connotations

oai_token-act-pair · gpt-3.5-turbo

references to biased perspectives and dimensions in discussions or arguments

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

osponsors

-0.87

ombs

-0.85

KEN

-0.84

gone

-0.80

eeks

-0.76

iques

-0.73

heon

-0.72

worthy

-0.72

TON

-0.71

KER

-0.71

POSITIVE LOGITS

sided

0.86

imensional

0.81

adoes

0.76

ide

0.67

 differe

0.67

ness

0.67

vantage

0.66

 Yamato

0.65

 plank

0.64

"{

0.63

Activations Density 0.030%

adjectives with negative connotations

references to biased perspectives and dimensions in discussions or arguments

No Comments

No Known Activations