INDEX

Explanations

organizations, locations, and events mentioned in news articles

oai_token-act-pair · gpt-3.5-turbo

numerical values and indicators of severity 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 @ 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

ichick

-0.86

emort

-0.82

 newsp

-0.76

Newsletter

-0.74

 utilizing

-0.67

catentry

-0.65

 favors

-0.65

 favoring

-0.64

 maneu

-0.63

 tremend

-0.62

POSITIVE LOGITS

BBC

0.97

Britain

0.83

Scientists

0.83

 Britain

0.77

British

0.77

 Dozens

0.77

 Britons

0.76

Hundreds

0.73

ÂłÂłÂł

0.73

 Rural

0.72

Activations Density 0.051%

organizations, locations, and events mentioned in news articles

numerical values and indicators of severity in various contexts

No Comments

No Known Activations