INDEX

Explanations

specific characters to accurately detect text sections

oai_token-act-pair · gpt-3.5-turbo

sentence-ending punctuation marks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 flagship

-0.85

 ecosystem

-0.83

 landscape

-0.81

 welcome

-0.81

 extinct

-0.79

omics

-0.77

 obsolete

-0.75

 ecosystems

-0.73

 capitals

-0.73

 interstellar

-0.72

POSITIVE LOGITS

 Investigators

1.40

 Prosecutors

1.31

 According

1.22

 However

1.22

 Afterwards

1.20

Attempts

1.17

 Later

1.16

 Then

1.16

 Initially

1.16

 Shortly

1.15

Activations Density 0.503%

specific characters to accurately detect text sections

sentence-ending punctuation marks

No Comments

No Known Activations