INDEX

Explanations

mentions of specific United States Senators

oai_token-act-pair · gpt-3.5-turbo

references to senators and congressional leadership

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

variable

-0.75

resp

-0.69

 Britann

-0.69

 Seed

-0.67

 Virus

-0.66

ology

-0.65

seed

-0.64

olon

-0.64

ological

-0.64

Dek

-0.64

POSITIVE LOGITS

 senators

1.05

Senate

0.95

iciary

0.92

 filib

0.90

¬¼

0.89

Republicans

0.89

 lawmakers

0.86

 Appropriations

0.85

 Republicans

0.85

ileaks

0.84

Activations Density 0.084%

mentions of specific United States Senators

references to senators and congressional leadership

No Comments

No Known Activations