INDEX

Explanations

links to share a specific story

oai_token-act-pair · gpt-3.5-turbo

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

quartered

-0.73

©¶æ¥µ

-0.66

¨

-0.65

ĵĺ

-0.63

pecially

-0.62

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.62

quarters

-0.61

 havens

-0.59

ãĤ©

-0.59

¹

-0.59

POSITIVE LOGITS

 story

1.19

 article

1.04

 STORY

0.88

 Story

0.85

 Article

0.79

article

0.79

 ARTICLE

0.72

 Neuroscience

0.71

 tale

0.71

Story

0.70

Activations Density 0.017%

links to share a specific story

No Comments

No Known Activations