INDEX

Explanations

events or actions that occurred in the past

oai_token-act-pair · gpt-3.5-turbo

instances of the word "earlier" indicating past events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

alion

-0.73

hop

-0.72

drivers

-0.68

lua

-0.66

pak

-0.65

ickle

-0.63

odan

-0.63

tre

-0.62

HO

-0.62

eric

-0.61

POSITIVE LOGITS

 versions

0.83

foundland

0.81

 generations

0.79

 than

0.79

 stages

0.74

than

0.74

 iterations

0.70

 eleph

0.69

 installments

0.69

 editions

0.68

Activations Density 0.025%

events or actions that occurred in the past

instances of the word "earlier" indicating past events

No Comments

No Known Activations