INDEX

Explanations

abbreviations followed by numbers

oai_token-act-pair · gpt-3.5-turbo

elements of humor and sarcasm

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

aditional

-0.85

 exha

-0.81

 undermin

-0.79

 eleph

-0.77

oun

-0.77

 proport

-0.71

 newcom

-0.71

 citiz

-0.70

 tremend

-0.69

senal

-0.68

POSITIVE LOGITS

Scroll

1.04

PHOTOS

0.97

Actor

0.78

Via

0.78

↵

0.78

SPONSORED

0.75

Els

0.74

Rum

0.73

lol

0.73

NPR

0.71

Activations Density 0.697%

abbreviations followed by numbers

elements of humor and sarcasm

No Comments

No Known Activations