INDEX

Explanations

terms related to cancer

oai_token-act-pair · gpt-3.5-turbo

references to cancer and related institutions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

worldly

-0.74

 Polk

-0.71

nee

-0.69

SPONSORED

-0.67

peg

-0.66

âĢ¢âĢ¢âĢ¢âĢ¢

-0.66

ances

-0.62

Drag

-0.62

TRUMP

-0.62

ele

-0.60

POSITIVE LOGITS

 Cancer

1.25

cancer

1.02

 cancer

1.01

 screenings

0.86

 Moons

0.83

alli

0.79

 cancers

0.78

oma

0.77

 metast

0.75

 carcin

0.75

Activations Density 0.005%

terms related to cancer

references to cancer and related institutions

No Comments

No Known Activations