INDEX

Explanations

news-related keywords and terms

oai_token-act-pair · gpt-3.5-turbo

references to the most significant news headlines or stories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Ã©n

-0.76

 Quart

-0.74

amiya

-0.67

due

-0.66

livious

-0.65

izoph

-0.63

gged

-0.63

oler

-0.63

âĸĪâĸĪâĸĪâĸĪ

-0.62

aque

-0.62

POSITIVE LOGITS

 updates

0.97

 news

0.94

 bang

0.92

 headlines

0.90

 scoop

0.89

 discounts

0.88

 NEWS

0.80

 rugby

0.79

 savings

0.77

 benefits

0.76

Activations Density 0.023%

news-related keywords and terms

references to the most significant news headlines or stories

No Comments

No Known Activations