INDEX

Explanations

questions asking for reasons or explanations

oai_token-act-pair · gpt-3.5-turbo

questions and phrases involving uncertainty or conditional scenarios

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

://

-0.60

geon

-0.58

SourceFile

-0.57

 sidelines

-0.56

stroke

-0.56

notations

-0.55

 veterans

-0.53

fax

-0.53

 prints

-0.53

PORT

-0.53

POSITIVE LOGITS

uh

0.77

yes

0.75

 anyway

0.74

 somew

0.72

WHY

0.70

 exactly

0.69

?)

0.67

 pray

0.66

ileged

0.66

oulos

0.66

Activations Density 0.107%

questions asking for reasons or explanations

questions and phrases involving uncertainty or conditional scenarios

No Comments

No Known Activations