INDEX

Explanations

words related to current events or current statuses

oai_token-act-pair · gpt-3.5-turbo

the word "current" and its variations, indicating a focus on ongoing or recent subjects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

Bei

-0.90

alty

-0.86

romeda

-0.85

wich

-0.79

jah

-0.79

alties

-0.78

essen

-0.76

phasis

-0.76

otle

-0.73

ambo

-0.72

POSITIVE LOGITS

 incarnation

0.84

 whereabouts

0.80

 circumstances

0.78

 predicament

0.76

 affairs

0.70

 occupant

0.70

URRENT

0.69

 resident

0.68

 holders

0.66

 trends

0.65

Activations Density 0.019%

words related to current events or current statuses

the word "current" and its variations, indicating a focus on ongoing or recent subjects

No Comments

No Known Activations