INDEX

Explanations

names and terms related to persons or organizations

oai_token-act-pair · gpt-3.5-turbo

specific names or titles, particularly those related to individuals or authorities

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

 Tycoon

-0.73

 Scotia

-0.72

 Canaver

-0.69

 GEAR

-0.66

 saline

-0.65

 Durham

-0.62

WHERE

-0.62

 Makoto

-0.61

 custody

-0.60

âĢ¢âĢ¢âĢ¢âĢ¢

-0.59

POSITIVE LOGITS

wered

1.42

Ans

1.10

linger

1.06

ensibly

0.94

ensible

0.90

ovic

0.88

entials

0.88

pread

0.88

combe

0.87

anta

0.87

Activations Density 0.018%

names and terms related to persons or organizations

specific names or titles, particularly those related to individuals or authorities

No Comments

No Known Activations