INDEX

Explanations

words related to greetings and wishes

oai_token-act-pair · gpt-3.5-turbo

specific names or titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 presumption

-0.61

 applicable

-0.61

 deduction

-0.57

 revocation

-0.56

iaries

-0.56

 condem

-0.55

 imagination

-0.55

 uninsured

-0.55

 ______

-0.54

 unmarried

-0.53

POSITIVE LOGITS

Ã±o

1.18

Corp

1.01

Ã±a

1.00

ÅĤ

0.99

uthor

0.98

Ã±

0.96

hedral

0.94

qua

0.93

Wan

0.93

Lens

0.91

Activations Density 0.333%

words related to greetings and wishes

specific names or titles

No Comments

No Known Activations