INDEX

Explanations

names of individuals, possibly notable personalities

oai_token-act-pair · gpt-3.5-turbo

occurrences of specific names or initials related to individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 envy

-0.62

 Shiv

-0.62

 shorthand

-0.57

SHIP

-0.55

 Boxing

-0.55

vs

-0.55

pedia

-0.54

 Murdoch

-0.54

ä»

-0.51

 gigg

-0.51

POSITIVE LOGITS

ologne

0.70

ĸļ

0.69

roma

0.67

eret

0.67

uren

0.65

tera

0.62

irable

0.62

isi

0.62

igne

0.61

escal

0.61

Activations Density 0.073%

names of individuals, possibly notable personalities

occurrences of specific names or initials related to individuals

No Comments

No Known Activations