INDEX

Explanations

phrases related to entries or enrollments

oai_token-act-pair · gpt-3.5-turbo

references to entry points or instances in a context involving selection or admission

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Pis

-0.73

 Dist

-0.69

PBS

-0.66

 pink

-0.64

Bit

-0.61

 film

-0.61

 bleach

-0.61

 fund

-0.60

 Pink

-0.60

 sexually

-0.59

POSITIVE LOGITS

entry

4.83

Entry

1.92

 Entry

1.74

 entry

1.48

enter

1.39

ent

1.19

Ent

0.97

door

0.96

 entries

0.94

ittance

0.92

Activations Density 0.011%

phrases related to entries or enrollments

references to entry points or instances in a context involving selection or admission

No Comments

No Known Activations