INDEX

Explanations

names of individuals, especially with titles like "Dr." and "Mrs."

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 raft

-0.93

NAT

-0.86

pol

-0.86

Pol

-0.85

ÅĤ

-0.83

Magikarp

-0.80

 Slav

-0.78

oj

-0.77

ossal

-0.77

Ü

-0.77

POSITIVE LOGITS

 Davis

2.40

Davis

2.30

 Davies

1.24

 Hayes

1.19

avis

1.12

 Ellis

1.09

 Dean

1.07

 Curtis

1.03

 Rowe

1.03

 Luke

1.03

Activations Density 0.247%

names of individuals, especially with titles like "Dr." and "Mrs."

No Comments

No Known Activations