INDEX

Explanations

phrases related to both sides of an argument or issue

oai_token-act-pair · gpt-3.5-turbo

phrases related to political discourse and division

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Pwr

-0.81

pta

-0.68

orable

-0.66

nce

-0.65

nces

-0.64

eday

-0.62

ements

-0.61

'>

-0.61

RAM

-0.60

\<

-0.60

POSITIVE LOGITS

 aisle

1.32

 spectrum

0.97

 fence

0.96

 coin

0.93

 ledger

0.87

 Atlantic

0.80

 equation

0.79

 pond

0.78

 fences

0.77

 divide

0.77

Activations Density 0.089%

phrases related to both sides of an argument or issue

phrases related to political discourse and division

No Comments

No Known Activations