INDEX

Explanations

strings related to names and titles in various contexts

oai_token-act-pair · gpt-3.5-turbo

names of public figures and notable individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 GOODMAN

-0.59

xiety

-0.55

isks

-0.54

 deterrent

-0.54

metics

-0.53

 Inventory

-0.52

 âĢº

-0.51

Reviewer

-0.49

 NETWORK

-0.49

ependent

-0.49

POSITIVE LOGITS

 himself

0.80

itone

0.68

anky

0.66

çĶ°

0.65

 assassinated

0.63

éĥ

0.60

his

0.59

ãĥīãĥ©ãĤ´ãĥ³

0.59

etter

0.58

selage

0.57

Activations Density 2.015%

strings related to names and titles in various contexts

names of public figures and notable individuals

No Comments

No Known Activations