INDEX

Explanations

references to violence and destruction

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "ripped" and its variations

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

 Modes

-0.71

ithmetic

-0.68

Alm

-0.67

stay

-0.65

 Factors

-0.64

AI

-0.64

Ind

-0.63

 Gallup

-0.62

Ba

-0.62

subject

-0.62

POSITIVE LOGITS

 ripped

3.48

 ripping

2.26

 tore

1.93

 shredded

1.65

 torn

1.62

rip

1.55

 blasted

1.39

 stripped

1.36

 slammed

1.33

 snatched

1.32

Activations Density 0.012%

references to violence and destruction

occurrences of the word "ripped" and its variations

No Comments

No Known Activations