INDEX

Explanations

content related to unofficial or official status

oai_token-act-pair · gpt-3.5-turbo

terms related to official sources or documentation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 posture

-0.69

 expiration

-0.68

 array

-0.67

 knees

-0.67

 model

-0.67

 beetles

-0.67

lob

-0.67

 options

-0.66

 plants

-0.66

 compassion

-0.66

POSITIVE LOGITS

official

4.16

Official

2.11

offic

1.62

natural

1.27

orthodox

1.22

popular

1.19

usual

1.13

famous

1.11

regular

1.08

latest

1.06

Activations Density 0.034%

content related to unofficial or official status

terms related to official sources or documentation

No Comments

No Known Activations