INDEX

Explanations

news-related words and phrases

oai_token-act-pair · gpt-3.5-turbo

references to breaking news and news alerts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ignt

-0.75

ridge

-0.75

 grun

-0.72

ued

-0.71

arer

-0.67

itent

-0.67

 SOLD

-0.67

ngth

-0.66

lac

-0.66

inka

-0.65

POSITIVE LOGITS

reader

0.88

ource

0.82

Catholic

0.80

Trend

0.80

 headlines

0.80

 Tonight

0.78

Reporting

0.76

flash

0.76

worthy

0.76

NEWS

0.74

Activations Density 0.032%

news-related words and phrases

references to breaking news and news alerts

No Comments

No Known Activations