INDEX

Explanations

references to the frequency or manner in which something is used

oai_token-act-pair · gpt-3.5-turbo

terms related to the frequency or amount of use

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

inosaur

-0.77

tell

-0.70

hair

-0.65

meat

-0.64

 Barg

-0.62

 naked

-0.62

bitious

-0.60

 Drac

-0.60

 Kore

-0.59

abella

-0.59

POSITIVE LOGITS

 habits

0.95

ername

0.92

iences

0.84

 patterns

0.75

ickr

0.72

 interfaces

0.71

ately

0.70

notations

0.70

TOP

0.67

fulness

0.66

Activations Density 0.021%

references to the frequency or manner in which something is used

terms related to the frequency or amount of use

No Comments

No Known Activations