INDEX

Explanations

words related to starting or initiating actions or processes

oai_token-act-pair · gpt-3.5-turbo

phrases related to initiation and involvement in activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 ambush

-0.60

 Methods

-0.58

 entirety

-0.57

 former

-0.55

atton

-0.55

 retired

-0.55

chief

-0.55

aut

-0.54

atern

-0.53

lf

-0.53

POSITIVE LOGITS

ãĤ¼

0.83

retty

0.82

ocobo

0.82

Reloaded

0.73

chy

0.72

 traction

0.72

ipolar

0.68

started

0.67

actionDate

0.66

 sidx

0.66

Activations Density 0.248%

words related to starting or initiating actions or processes

phrases related to initiation and involvement in activities

No Comments

No Known Activations