INDEX

Explanations

phrases indicating sequence or order

oai_token-act-pair · gpt-3.5-turbo

instances of the word "then" used to indicate sequential actions or events

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

 belief

-0.64

 impossibility

-0.63

 Presence

-0.61

 Doctrine

-0.61

toe

-0.59

Bie

-0.58

 competence

-0.58

 Transactions

-0.58

ones

-0.57

 asthma

-0.57

POSITIVE LOGITS

 proceeded

1.34

 promptly

0.92

 proceed

0.89

 proceeds

0.89

 transitioned

0.86

forth

0.82

 swiftly

0.78

 embarked

0.78

nell

0.78

 followed

0.75

Activations Density 0.055%

phrases indicating sequence or order

instances of the word "then" used to indicate sequential actions or events

No Comments

No Known Activations