INDEX

Explanations

political affiliations, especially focusing on identifying individuals as Republicans or Democrats

oai_token-act-pair · gpt-3.5-turbo

references to political party affiliations and actions, particularly related to Republicans and Democrats

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

ysc

-0.78

 Afric

-0.78

sis

-0.73

bred

-0.72

zsche

-0.72

 Seym

-0.72

weather

-0.71

worthiness

-0.68

tics

-0.68

adr

-0.67

POSITIVE LOGITS

 lawmakers

0.91

 senators

0.90

 majorities

0.84

 congressional

0.84

 lawmaker

0.81

 legislators

0.79

 stal

0.78

 Republicans

0.73

 filib

0.72

 insistence

0.70

Activations Density 0.092%

political affiliations, especially focusing on identifying individuals as Republicans or Democrats

references to political party affiliations and actions, particularly related to Republicans and Democrats

No Comments

No Known Activations