INDEX

Explanations

phrases related to completing tasks or processes, often related to work or productivity

oai_token-act-pair · gpt-3.5-turbo

instances of communication or calls in a social context

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

 Zimbabwe

-0.68

ilial

-0.64

andestine

-0.62

anwhile

-0.61

 mistrust

-0.60

asha

-0.59

seek

-0.59

overe

-0.58

 hierarch

-0.58

 hierarchy

-0.57

POSITIVE LOGITS

 mileage

0.86

 traction

0.82

OH

0.69

Rid

0.68

 bang

0.68

 permission

0.67

 closer

0.65

DERR

0.64

ãĥĸ

0.62

gran

0.62

Activations Density 0.393%

phrases related to completing tasks or processes, often related to work or productivity

instances of communication or calls in a social context

No Comments

No Known Activations