INDEX

Explanations

terms related to archiving or storing data

oai_token-act-pair · gpt-3.5-turbo

references to digital or physical archives

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

laws

-0.76

nir

-0.74

osuke

-0.73

ggle

-0.70

uana

-0.68

acted

-0.68

stad

-0.67

hua

-0.67

nel

-0.67

cker

-0.67

POSITIVE LOGITS

 archives

0.97

 trove

0.91

 Archive

0.89

 archive

0.89

 Archives

0.87

conservancy

0.86

 dumps

0.84

archives

0.83

 documenting

0.80

 collection

0.80

Activations Density 0.024%

terms related to archiving or storing data

references to digital or physical archives

No Comments

No Known Activations