INDEX

Explanations

proper nouns of news sources and authors

oai_token-act-pair · gpt-3.5-turbo

references to sources that provide commentary or analysis

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

miss

-0.54

 bapt

-0.53

artifacts

-0.53

 Predators

-0.53

mediated

-0.51

iets

-0.51

pillar

-0.50

abases

-0.50

ococ

-0.49

Lug

-0.49

POSITIVE LOGITS

(),

0.83

 elsewhere

0.82

 shortly

0.75

:[

0.74

 omin

0.73

,,

0.70

 eloqu

0.70

 recently

0.69

,.

0.69

0.68

Activations Density 0.074%

proper nouns of news sources and authors

references to sources that provide commentary or analysis

No Comments

No Known Activations