INDEX

Explanations

suggestions of wrongdoing or responsibility

oai_token-act-pair · gpt-3.5-turbo

references to individuals or entities responsible for certain actions or events

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

ISTER

-0.85

pec

-0.74

ORT

-0.72

cyclop

-0.68

formed

-0.67

©¶æ

-0.66

ort

-0.66

 funeral

-0.66

 Parables

-0.65

forms

-0.65

POSITIVE LOGITS

 culprit

1.11

0.98

ãĥ¼ãĥĨãĤ£

0.90

sie

0.86

icons

0.84

æ©

0.82

ãĤ¸

0.78

eous

0.78

DragonMagazine

0.75

enary

0.74

Activations Density 0.022%

suggestions of wrongdoing or responsibility

references to individuals or entities responsible for certain actions or events

No Comments

No Known Activations