INDEX

Explanations

negative words or phrases describing something as very bad or catastrophic

oai_token-act-pair · gpt-3.5-turbo

words and phrases that convey extreme negative or positive evaluations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ravings

-0.84

ancies

-0.83

 administrations

-0.78

ensions

-0.77

©¶æ

-0.76

tones

-0.74

went

-0.71

ipation

-0.69

 grievances

-0.68

rants

-0.68

POSITIVE LOGITS

 magnet

0.83

 jewel

0.80

 asset

0.80

 notch

0.77

oly

0.75

 opener

0.71

 Clicker

0.71

gem

0.71

 example

0.71

 boon

0.70

Activations Density 0.218%

negative words or phrases describing something as very bad or catastrophic

words and phrases that convey extreme negative or positive evaluations

No Comments

No Known Activations