INDEX

Explanations

proper names, particularly the name "Vincent"

oai_token-act-pair · gpt-3.5-turbo

references to the name "Vincent."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itbart

-0.74

Edited

-0.70

bell

-0.70

TPS

-0.69

adesh

-0.68

MER

-0.68

outed

-0.65

agnetic

-0.65

cffffcc

-0.65

anooga

-0.64

POSITIVE LOGITS

 Vincent

0.95

oso

0.91

 Scully

0.86

rette

0.79

 Vaughn

0.76

 Leone

0.75

 Townsend

0.75

 Observatory

0.73

 Vega

0.72

oshenko

0.71

Activations Density 0.012%

proper names, particularly the name "Vincent"

references to the name "Vincent."

No Comments

No Known Activations