INDEX

Explanations

words related to relationships or partnerships

oai_token-act-pair · gpt-3.5-turbo

references to companionship or romantic partnerships

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uman

-0.73

uther

-0.73

icum

-0.71

athered

-0.67

ights

-0.66

english

-0.65

undred

-0.65

Syrian

-0.64

ERROR

-0.64

eye

-0.64

POSITIVE LOGITS

 mate

1.35

 Mate

1.09

 mates

1.08

mate

0.79

daq

0.74

volent

0.73

 Amir

0.73

 Nieto

0.72

||||

0.71

 Markus

0.70

Activations Density 0.006%

words related to relationships or partnerships

references to companionship or romantic partnerships

No Comments

No Known Activations