INDEX

Explanations

occurrences of specific numbered years

oai_token-act-pair · gpt-3.5-turbo

references to specific numerical years in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

zsche

-0.87

ulkan

-0.82

tnc

-0.76

UTF

-0.74

ilitary

-0.73

achine

-0.72

 Cosponsors

-0.69

acus

-0.69

jri

-0.68

abilia

-0.68

POSITIVE LOGITS

 anniversary

0.75

nings

0.71

Initialized

0.68

ear

0.68

orate

0.64

 Onion

0.63

eve

0.62

 olds

0.62

veland

0.60

ULTS

0.60

Activations Density 0.078%

occurrences of specific numbered years

references to specific numerical years in various contexts

No Comments

No Known Activations