INDEX

Explanations

phrases or words related to correctness, suitability, or appropriateness

oai_token-act-pair · gpt-3.5-turbo

references to the right choices or appropriate actions in various contexts

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

sung

-0.76

NetMessage

-0.74

krit

-0.71

hern

-0.70

limited

-0.70

doms

-0.70

jay

-0.69

imentary

-0.68

famous

-0.65

argon

-0.65

POSITIVE LOGITS

 Cooke

0.67

 Scand

0.66

 Danish

0.63

Ec

0.63

Aust

0.63

 dentist

0.60

icro

0.60

 Croatian

0.60

 rabbi

0.59

 Croat

0.59

Activations Density 0.111%

phrases or words related to correctness, suitability, or appropriateness

references to the right choices or appropriate actions in various contexts

No Comments

No Known Activations