INDEX

Explanations

phrases indicating extremity or intensity of action or opinion

oai_token-act-pair · gpt-3.5-turbo

phrases indicating actions or opinions that go to extremes or limits

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

itu

-0.77

rio

-0.72

odes

-0.70

 Puppet

-0.69

cyclopedia

-0.68

tu

-0.66

otten

-0.66

 Pend

-0.64

ĭ

-0.63

odd

-0.63

POSITIVE LOGITS

 lengths

0.77

 differently

0.76

 stride

0.71

 unnoticed

0.68

 embro

0.66

 persu

0.65

 step

0.64

 overr

0.62

HER

0.61

rug

0.60

Activations Density 0.039%

phrases indicating extremity or intensity of action or opinion

phrases indicating actions or opinions that go to extremes or limits

No Comments

No Known Activations