INDEX

Explanations

numbers and technical information within structured text

oai_token-act-pair · gpt-3.5-turbo

references to numerical values, symbols, or specific formatting patterns

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

itiveness

-0.89

natureconservancy

-0.74

manship

-0.69

thren

-0.69

trial

-0.68

 ACTIONS

-0.67

achine

-0.63

uously

-0.63

ĵĺ

-0.62

wcs

-0.62

POSITIVE LOGITS

ibrary

0.80

pora

0.80

 womb

0.77

ĸļ

0.72

¿½

0.70

etus

0.67

PsyNetMessage

0.67

 curv

0.67

anmar

0.67

hyde

0.65

Activations Density 1.099%

numbers and technical information within structured text

references to numerical values, symbols, or specific formatting patterns

No Comments

No Known Activations