INDEX

Explanations

adjectives related to high levels of stress or pressure

oai_token-act-pair · gpt-3.5-turbo

situations or contexts described as tense

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 resides

-0.73

Main

-0.66

Found

-0.63

 Purch

-0.62

 commissions

-0.62

[/

-0.61

Disk

-0.61

opl

-0.61

 Char

-0.60

 manufacturers

-0.59

POSITIVE LOGITS

 tense

4.15

 tension

1.56

 uneasy

1.31

 strained

1.25

 fraught

1.19

 tensions

1.16

cla

1.14

 cramped

1.13

 chilly

1.12

 turbulent

1.12

Activations Density 0.021%

adjectives related to high levels of stress or pressure

situations or contexts described as tense

No Comments

No Known Activations