INDEX

Explanations

phrases related to proof and verification

oai_token-act-pair · gpt-3.5-turbo

phrases related to types of proof or evidence in arguments

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.81

netflix

-0.73

reddits

-0.70

adelphia

-0.68

 Rivals

-0.66

erest

-0.63

scape

-0.62

 toget

-0.62

anon

-0.60

rella

-0.60

POSITIVE LOGITS

concept

0.85

 concept

0.75

 Concept

0.73

 existence

0.70

 principle

0.69

arching

0.67

shit

0.65

 innocence

0.65

 occupancy

0.64

 pudding

0.64

Activations Density 0.067%

phrases related to proof and verification

phrases related to types of proof or evidence in arguments

No Comments

No Known Activations