INDEX

Explanations

proper nouns with hyphens, likely related to political figures or entities

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

names related to political figures or authorities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

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

 Stream

-0.94

 multipl

-0.91

 Pengu

-0.85

 Curt

-0.79

POV

-0.78

 Parad

-0.78

 Ampl

-0.75

DOT

-0.74

 Versus

-0.74

 Morrow

-0.74

POSITIVE LOGITS

sung

1.32

Il

1.23

bian

1.20

tha

1.20

Wan

1.20

Hy

1.18

hung

1.17

kee

1.16

Gy

1.14

Hun

1.14

Activations Density 0.035%

proper nouns with hyphens, likely related to political figures or entities

names related to political figures or authorities

No Comments

No Known Activations