INDEX

Explanations

paragraphs in news articles with proper names and locations, marked by a special character sequence 'Ċ'.

oai_token-act-pair · gpt-3.5-turbo

occurrences of structured reports or official statements

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

lihood

-0.64

 libertarians

-0.63

 hindsight

-0.60

frames

-0.59

 agility

-0.58

tics

-0.58

 obscurity

-0.57

 offline

-0.57

 inflation

-0.57

 deflation

-0.57

POSITIVE LOGITS

Interested

0.86

WASHINGTON

0.77

Speaking

0.76

Specifically

0.75

ç«

0.72

neau

0.71

Rh

0.71

Amid

0.70

æĪ¦

0.69

Led

0.69

Activations Density 0.176%

paragraphs in news articles with proper names and locations, marked by a special character sequence 'Ċ'.

occurrences of structured reports or official statements

No Comments

No Known Activations