INDEX

Explanations

names of individuals

oai_token-act-pair · gpt-3.5-turbo

word fragments or syllables, particularly those related to names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 slee

-0.60

PSU

-0.59

ournals

-0.58

 Morty

-0.57

abee

-0.55

ultural

-0.54

NCT

-0.53

awed

-0.53

intosh

-0.52

)]

-0.52

POSITIVE LOGITS

imil

0.87

ãĤ´ãĥ³

0.73

illes

0.72

ĪĴ

0.72

ILLE

0.72

INAL

0.71

Ë

0.69

Ŀ

0.68

ħĭ

0.67

¥µ

0.66

Activations Density 0.092%

names of individuals

word fragments or syllables, particularly those related to names

No Comments

No Known Activations