INDEX

Explanations

references to a specific organization or event, potentially related to a forum or conference

oai_token-act-pair · gpt-3.5-turbo

the repeated use of the word "WE."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ph

-0.63

 stops

-0.63

 laps

-0.62

Rom

-0.61

 photography

-0.61

 Inqu

-0.60

 notes

-0.58

 assistant

-0.57

 another

-0.57

 Card

-0.57

POSITIVE LOGITS

WE

4.16

WE

1.75

we

1.53

HE

1.42

WH

1.35

WOOD

1.21

JO

1.18

YOU

1.17

BE

1.15

WHO

1.11

Activations Density 0.008%

references to a specific organization or event, potentially related to a forum or conference

the repeated use of the word "WE."

No Comments

No Known Activations