INDEX

Explanations

years mentioned within the early to mid-2000s

oai_token-act-pair · gpt-3.5-turbo

references to specific years in the early 2000s

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

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

place

-0.73

 clipboard

-0.60

nces

-0.57

akable

-0.56

 stead

-0.55

este

-0.54

places

-0.54

queue

-0.54

ampunk

-0.53

anch

-0.52

POSITIVE LOGITS

-'

1.09

å¹

0.82

ãĥĥãĥī

0.74

âĢ²

0.71

 vintage

0.69

–

0.69

 onwards

0.68

ãĤ¦ãĤ¹

0.65

âĢĳ

0.63

iyah

0.62

Activations Density 0.078%

years mentioned within the early to mid-2000s

references to specific years in the early 2000s

No Comments

No Known Activations