INDEX

Explanations

phrases signaling information attribution or citation

oai_token-act-pair · gpt-3.5-turbo

the word "According" used in various contexts, indicating references or attributions in sentences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

-0.81

 breeding

-0.66

 ashore

-0.64

 spiders

-0.64

 farm

-0.63

ãĥĵ

-0.63

ãĥŁ

-0.63

andering

-0.62

 patience

-0.62

 retreat

-0.61

POSITIVE LOGITS

 Sources

0.89

chwitz

0.88

According

0.87

rary

0.84

ety

0.81

ccording

0.81

glomer

0.79

tesy

0.77

Style

0.77

Format

0.76

Activations Density 0.014%

phrases signaling information attribution or citation

the word "According" used in various contexts, indicating references or attributions in sentences

No Comments

No Known Activations