INDEX

Explanations

phrases indicating a causative relationship or specific timing

oai_token-act-pair · gpt-3.5-turbo

the word "when" indicating temporal context

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

agin

-0.69

harm

-0.67

gem

-0.65

rolet

-0.63

Die

-0.62

Bas

-0.60

yre

-0.60

hou

-0.60

number

-0.60

nut

-0.59

POSITIVE LOGITS

soever

1.39

*/(

0.76

irlf

0.71

 confronted

0.70

theless

0.68

 asked

0.68

phies

0.66

etheless

0.65

 pressed

0.64

 faced

0.64

Activations Density 0.097%

phrases indicating a causative relationship or specific timing

the word "when" indicating temporal context

No Comments

No Known Activations