INDEX

Explanations

mentions of serious accusations or claims

oai_token-act-pair · gpt-3.5-turbo

references to allegations in various contexts

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

emaker

-0.69

 wisely

-0.67

iol

-0.67

 Plasma

-0.65

ete

-0.65

iet

-0.63

 saved

-0.63

 ðŁĻĤ

-0.63

 tuned

-0.63

arius

-0.63

POSITIVE LOGITS

 allegations

3.43

 accusations

2.85

 allegation

2.64

 accusation

2.10

 suspicions

1.87

 revelations

1.84

 assertions

1.83

 claims

1.79

 rumours

1.72

 complaints

1.70

Activations Density 0.020%

mentions of serious accusations or claims

references to allegations in various contexts

No Comments

No Known Activations