INDEX

Explanations

references to health conditions, especially with a focus on diseases

oai_token-act-pair · gpt-3.5-turbo

references to various diseases

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

endas

-0.79

ruce

-0.79

adra

-0.78

ramid

-0.77

ilts

-0.76

anian

-0.76

idav

-0.76

adesh

-0.75

ogle

-0.73

assy

-0.72

POSITIVE LOGITS

 disease

0.97

 outbreak

0.97

 Immun

0.96

 prevention

0.93

 diseases

0.90

 Outbreak

0.90

 outbreaks

0.88

 Disease

0.85

 Diseases

0.83

 plague

0.80

Activations Density 0.016%

references to health conditions, especially with a focus on diseases

references to various diseases

No Comments

No Known Activations