INDEX

Explanations

official websites or documents

oai_token-act-pair · gpt-3.5-turbo

references to official documents and sources

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

Interstitial

-0.69

ively

-0.67

 randomly

-0.65

stakes

-0.64

 Lomb

-0.64

 strategically

-0.63

entially

-0.63

 spat

-0.63

 centrif

-0.63

rily

-0.62

POSITIVE LOGITS

 corrid

0.91

ĳå£«

0.81

acea

0.76

SpaceEngineers

0.75

achable

0.74

escription

0.70

OOD

0.69

Initialized

0.69

incial

0.68

\\\\\\\\

0.66

Activations Density 0.192%

official websites or documents

references to official documents and sources

No Comments

No Known Activations