INDEX

Explanations

phrases related to returning to a previous state or condition

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

detail

-0.62

ebin

-0.62

 translating

-0.61

 translation

-0.59

ãĤ¼ãĤ¦ãĤ¹

-0.58

Translation

-0.58

PI

-0.58

CONCLUS

-0.57

illin

-0.57

fax

-0.57

POSITIVE LOGITS

 previously

1.46

 formerly

1.31

 previous

1.29

 originally

1.21

 earlier

1.14

 Previously

1.06

 predecessors

1.05

 Originally

1.05

 predecessor

1.04

 pioneered

0.99

Activations Density 0.410%

phrases related to returning to a previous state or condition

No Comments

No Known Activations