INDEX

Explanations

terms related to knowledge and knowing

oai_token-act-pair · gpt-3.5-turbo

references to knowledge, knowledge-related concepts, and understanding

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 replacement

-0.77

 è£ıç

-0.70

led

-0.68

Pop

-0.66

Fas

-0.65

 prop

-0.65

upl

-0.65

 unit

-0.64

spl

-0.64

 component

-0.64

POSITIVE LOGITS

know

4.22

Know

2.46

 KNOW

1.84

knowledge

1.77

 Know

1.58

seeing

1.37

Knowing

1.33

 know

1.33

remember

1.30

learn

1.24

Activations Density 0.012%

terms related to knowledge and knowing

references to knowledge, knowledge-related concepts, and understanding

No Comments

No Known Activations