INDEX

Explanations

phrases related to trustworthy news sources

oai_token-act-pair · gpt-3.5-turbo

questions and expressions of interest or engagement

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

hement

-0.78

 eroded

-0.76

hene

-0.72

isons

-0.69

 decap

-0.68

 reneg

-0.68

 estranged

-0.68

 unaccount

-0.68

acle

-0.66

 dismantled

-0.66

POSITIVE LOGITS

Disclaimer

0.95

âĺħ

0.90

âĿ

0.87

VOL

0.85

Brow

0.84

Click

0.84

Mouse

0.84

Ye

0.83

GU

0.82

Subscribe

0.82

Activations Density 0.257%

phrases related to trustworthy news sources

questions and expressions of interest or engagement

No Comments

No Known Activations