INDEX

Explanations

mentions of 'ultra' in a context that suggests uniqueness, intensity, or extremeness

oai_token-act-pair · gpt-3.5-turbo

frequent mentions of the term "ultra" in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Laugh

-0.82

 Shape

-0.80

 Burnett

-0.77

 Manifest

-0.73

 Bleach

-0.72

 Landing

-0.72

 Nights

-0.71

stocks

-0.71

 Quarter

-0.69

nce

-0.69

POSITIVE LOGITS

 ultra

1.19

orthodox

0.99

 ultr

0.95

avorite

0.89

entimes

0.83

yip

0.83

achev

0.81

thodox

0.81

visor

0.80

 heavyweight

0.79

Activations Density 0.006%

mentions of 'ultra' in a context that suggests uniqueness, intensity, or extremeness

frequent mentions of the term "ultra" in various contexts

No Comments

No Known Activations