INDEX

Explanations

phrases related to political events and entities

oai_token-act-pair · gpt-3.5-turbo

phrases or constructs involving "of" and "by," particularly in contexts related to allegations or accusations

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

 somew

-0.77

 vulner

-0.75

 somet

-0.69

pmwiki

-0.68

[&

-0.68

,—

-0.68

izoph

-0.68

BILITIES

-0.67

fuck

-0.64

Dispatch

-0.61

POSITIVE LOGITS

 President

1.08

 Barack

1.06

 disgr

0.97

 billionaire

0.89

 ousted

0.86

 former

0.82

 George

0.80

 presumptive

0.78

 Britain

0.77

 Donald

0.77

Activations Density 0.411%

phrases related to political events and entities

phrases or constructs involving "of" and "by," particularly in contexts related to allegations or accusations

No Comments

No Known Activations