INDEX

Explanations

proper names, specifically focusing on the name "Fernando"

oai_token-act-pair · gpt-3.5-turbo

references to individuals named Fernando, particularly in the context of sports or notable achievements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

oker

-0.84

prus

-0.82

henko

-0.76

wow

-0.74

iley

-0.74

icka

-0.72

ategory

-0.72

figured

-0.71

å¥³

-0.71

pell

-0.70

POSITIVE LOGITS

 Alonso

0.92

 Fernando

0.76

Oro

0.75

irect

0.75

 Talent

0.74

enne

0.72

 Rico

0.72

 NiÃ±o

0.72

iment

0.71

eering

0.71

Activations Density 0.022%

proper names, specifically focusing on the name "Fernando"

references to individuals named Fernando, particularly in the context of sports or notable achievements

No Comments

No Known Activations