INDEX

Explanations

words related to technology and technical terms, especially those starting with "E"

oai_token-act-pair · gpt-3.5-turbo

references to specific events or entities denoted by 'E' followed by numbers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Ashton

-0.67

 interviewing

-0.63

 instant

-0.61

Aph

-0.61

 sooner

-0.60

 Akron

-0.60

 excessively

-0.60

 names

-0.60

 Kenobi

-0.60

 voices

-0.60

POSITIVE LOGITS

cosystem

1.35

nerg

1.29

ighty

1.25

fficient

1.23

tymology

1.21

ASY

1.21

CHO

1.18

thereal

1.17

minent

1.15

rika

1.15

Activations Density 0.024%

words related to technology and technical terms, especially those starting with "E"

references to specific events or entities denoted by 'E' followed by numbers

No Comments

No Known Activations