INDEX

Explanations

proper nouns, specifically focusing on names of individuals and places

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 alarm

-0.71

 bias

-0.68

 account

-0.67

Hal

-0.62

 cutting

-0.61

cut

-0.59

 adjusting

-0.59

 estimation

-0.59

Im

-0.59

 research

-0.59

POSITIVE LOGITS

vez

5.32

lez

1.36

avez

1.31

alez

1.29

pez

1.14

rez

1.13

encia

1.01

 Chavez

0.97

urat

0.96

irez

0.95

Activations Density 0.018%

proper nouns, specifically focusing on names of individuals and places

proper nouns, specifically names

No Comments

No Known Activations