INDEX

Explanations

phrases indicating a directive or instruction

oai_token-act-pair · gpt-3.5-turbo

statements regarding personal rights or autonomy

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 guiActiveUnfocused

-0.79

ĺħ

-0.69

atro

-0.66

ascal

-0.65

ools

-0.65

oward

-0.64

assian

-0.61

ries

-0.61

Chair

-0.61

 chaired

-0.59

POSITIVE LOGITS

abouts

0.76

NESS

0.72

uni

0.72

Pyr

0.68

optional

0.67

ighting

0.66

Eps

0.65

pai

0.64

nz

0.64

ski

0.60

Activations Density 0.000%

phrases indicating a directive or instruction

statements regarding personal rights or autonomy

No Comments

No Known Activations