INDEX

Explanations

mentions of positive audience reactions like standing ovations

oai_token-act-pair · gpt-3.5-turbo

the word "standing" in various contexts and usages

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

ilers

-0.83

iler

-0.75

[|

-0.73

ocaly

-0.72

egu

-0.71

ffic

-0.71

rez

-0.71

esters

-0.70

xy

-0.70

killer

-0.69

POSITIVE LOGITS

still

1.06

 upright

0.87

 tall

0.77

 desks

0.74

 taller

0.71

 seiz

0.68

 steadfast

0.68

ov

0.68

anke

0.66

 unst

0.66

Activations Density 0.016%

mentions of positive audience reactions like standing ovations

the word "standing" in various contexts and usages

No Comments

No Known Activations