INDEX

Explanations

mentions of political parties, leadership, and party-related events

oai_token-act-pair · gpt-3.5-turbo

references to a political party or discussions related to political parties

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

 Wheat

-0.73

 Tile

-0.73

angelo

-0.73

alam

-0.71

 Deter

-0.71

 Drake

-0.71

 Grizz

-0.67

DERR

-0.67

Maw

-0.66

 Stafford

-0.66

POSITIVE LOGITS

leader

0.96

 affiliation

0.96

Leader

0.89

goers

0.87

 insiders

0.84

 affili

0.84

 leader

0.82

ratulations

0.78

 treasurer

0.77

agogue

0.77

Activations Density 0.031%

mentions of political parties, leadership, and party-related events

references to a political party or discussions related to political parties

No Comments

No Known Activations