INDEX

Explanations

conjunctions and prepositions

oai_token-act-pair · gpt-3.5-turbo

repeated conjunctions and phrases indicating association or connection

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

 tremend

-0.61

 eleph

-0.59

 earthqu

-0.59

 Dempsey

-0.56

 Tray

-0.53

 pione

-0.51

 Vance

-0.50

ģ«

-0.50

 Rider

-0.49

Þ

-0.49

POSITIVE LOGITS

1.58

-[

1.39

-$

1.27

âĢĳ

1.27

-)

1.21

'-

1.19

-(

1.17

-'

1.17

-.

1.15

-,

1.13

Activations Density 0.168%

conjunctions and prepositions

repeated conjunctions and phrases indicating association or connection

No Comments

No Known Activations