INDEX

Explanations

phrases expressing a lack of concern or importance

oai_token-act-pair · gpt-3.5-turbo

repeated phrases indicating the absence of wrongdoing or controversy

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

anos

-0.79

odan

-0.74

 Pastebin

-0.67

ilts

-0.67

bda

-0.67

orks

-0.65

 Hits

-0.65

eda

-0.63

eds

-0.63

ahime

-0.62

POSITIVE LOGITS

 else

1.10

 wrong

1.10

 stopping

1.01

 inherently

1.01

 intrinsically

0.91

 happening

0.89

 objectionable

0.83

 inherent

0.81

 intrinsic

0.81

 worse

0.81

Activations Density 0.059%

phrases expressing a lack of concern or importance

repeated phrases indicating the absence of wrongdoing or controversy

No Comments

No Known Activations