INDEX

Explanations

named entities and locations, specifically those with Eastern European origin

oai_token-act-pair · gpt-3.5-turbo

references to specific dates or chronological markers

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

 oats

-0.72

 casc

-0.69

 redund

-0.68

 Sturgeon

-0.65

 Replay

-0.63

onse

-0.62

aching

-0.62

erest

-0.62

 Scotia

-0.60

htaking

-0.60

POSITIVE LOGITS

ovic

1.01

ovi

0.96

jan

0.87

uary

0.86

ority

0.86

ners

0.84

estamp

0.82

owski

0.81

ovsky

0.81

ifier

0.79

Activations Density 0.028%

named entities and locations, specifically those with Eastern European origin

references to specific dates or chronological markers

No Comments

No Known Activations