INDEX

Explanations

words related to medical conditions, particularly those related to glands and cancers

oai_token-act-pair · gpt-3.5-turbo

medical terms related to cancer and hormone-related conditions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aneers

-0.88

ombie

-0.82

assian

-0.80

Dak

-0.73

Fey

-0.72

Mah

-0.70

arta

-0.69

erry

-0.68

ord

-0.68

 Doctrine

-0.67

POSITIVE LOGITS

 gland

1.30

 prostate

1.11

 cancer

1.01

 glands

0.95

 cancers

0.91

cancer

0.85

 Cancer

0.85

 ejac

0.81

tum

0.77

 calcul

0.76

Activations Density 0.009%

words related to medical conditions, particularly those related to glands and cancers

medical terms related to cancer and hormone-related conditions

No Comments

No Known Activations