INDEX

Explanations

names of individuals, possibly in a list or sequence

oai_token-act-pair · gpt-3.5-turbo

proper nouns, especially names and titles

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

bryce

-0.68

Pwr

-0.62

osterone

-0.61

 Roundup

-0.60

 Skydragon

-0.59

 captivity

-0.57

 disson

-0.55

 yourselves

-0.54

 prost

-0.54

 plaque

-0.53

POSITIVE LOGITS

ilver

0.83

vu

0.80

hua

0.72

inen

0.72

edu

0.67

direction

0.67

zee

0.65

ilan

0.64

vre

0.64

fecture

0.64

Activations Density 0.145%

names of individuals, possibly in a list or sequence

proper nouns, especially names and titles

No Comments

No Known Activations