INDEX

Explanations

governmental terms related to political administrations

oai_token-act-pair · gpt-3.5-turbo

references to government administrations, particularly the Obama and Trump administrations

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

actionDate

-0.73

 Haram

-0.70

Ward

-0.70

Crime

-0.68

#$#$

-0.68

Neigh

-0.67

Kin

-0.67

Pitt

-0.66

Saint

-0.66

aceous

-0.66

POSITIVE LOGITS

 appoint

0.80

 secretaries

0.80

 nominated

0.79

 shutdown

0.78

 officials

0.78

 appointed

0.77

 tasked

0.75

 motto

0.75

 bureaucracy

0.75

ahime

0.74

Activations Density 0.024%

governmental terms related to political administrations

references to government administrations, particularly the Obama and Trump administrations

No Comments

No Known Activations