INDEX

Explanations

phrases indicating information or instructions given to others

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "we were told."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

Labor

-0.74

adesh

-0.68

ouf

-0.68

hur

-0.65

cession

-0.65

aband

-0.64

ent

-0.63

aho

-0.63

avez

-0.61

ivot

-0.61

POSITIVE LOGITS

tale

0.80

llor

0.72

ariat

0.71

ÃĽ

0.68

 repeatedly

0.68

 perspect

0.67

ļé

0.67

 proced

0.65

Īè

0.65

 aback

0.65

Activations Density 0.027%

phrases indicating information or instructions given to others

instances of the phrase "we were told."

No Comments

No Known Activations