INDEX

Explanations

phrases related to investigations or criminal activities

oai_token-act-pair · gpt-3.5-turbo

references to investigations and claims involving specific entities or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 whatsoever

-0.81

 gladly

-0.81

 anytime

-0.79

 besides

-0.75

 sooner

-0.75

respective

-0.74

 importantly

-0.73

 everywhere

-0.71

Any

-0.71

 fullest

-0.70

POSITIVE LOGITS

an

0.93

 another

0.93

0.91

 Frenchman

0.80

 hers

0.80

 Franz

0.77

 Moroccan

0.75

 Ethiopian

0.74

Moz

0.74

 Saddam

0.74

Activations Density 0.478%

phrases related to investigations or criminal activities

references to investigations and claims involving specific entities or conditions

No Comments

No Known Activations