INDEX

Explanations

contractions where the letters 'a' and 'o' have been replaced with special characters like âĢ and Ļ

oai_token-act-pair · gpt-3.5-turbo

instances of the word "haven" in various contexts, particularly its negative form

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 manifests

-0.76

cision

-0.67

reme

-0.66

cuts

-0.65

dress

-0.64

ugu

-0.62

BD

-0.61

rophe

-0.61

oral

-0.59

rop

-0.59

POSITIVE LOGITS

't

1.12

geon

0.91

 Been

0.88

vier

0.81

ajor

0.80

nas

0.80

Wem

0.79

GOT

0.78

ited

0.76

ÃŃ

0.75

Activations Density 0.008%

contractions where the letters 'a' and 'o' have been replaced with special characters like âĢ and Ļ

instances of the word "haven" in various contexts, particularly its negative form

No Comments

No Known Activations