INDEX

Explanations

written content that appeared in different mediums or platforms

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the publication or origin of articles and stories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 nonetheless

-0.78

respective

-0.70

 constantly

-0.69

verend

-0.67

meric

-0.65

urrent

-0.64

ĺħ

-0.64

 preferably

-0.63

mediate

-0.63

 invariably

-0.63

POSITIVE LOGITS

arthed

0.71

HAM

0.64

 Dunham

0.63

Eag

0.62

Te

0.60

doi

0.60

orthy

0.59

ËĪ

0.58

 Canaveral

0.58

APH

0.58

Activations Density 0.163%

written content that appeared in different mediums or platforms

phrases indicating the publication or origin of articles and stories

No Comments

No Known Activations