INDEX

Explanations

publication-related terms, likely related to journalistic articles and posts

oai_token-act-pair · gpt-3.5-turbo

instances of the term "posted" and other related publication timestamps

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

rium

-0.78

tools

-0.67

abouts

-0.67

ilitarian

-0.66

cases

-0.65

finder

-0.64

anca

-0.63

atever

-0.62

igans

-0.62

gui

-0.61

POSITIVE LOGITS

Tue

0.80

Tue

0.79

Thu

0.71

 July

0.70

 February

0.69

 October

0.69

Published

0.69

 March

0.68

 January

0.68

Thu

0.65

Activations Density 0.054%

publication-related terms, likely related to journalistic articles and posts

instances of the term "posted" and other related publication timestamps

No Comments

No Known Activations