INDEX

Explanations

titles or phrases that indicate the beginning of an idea or topic

oai_token-act-pair · gpt-3.5-turbo

punctuation and question marks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

roud

-0.71

apsed

-0.65

éĹĺ

-0.65

 earthqu

-0.61

itton

-0.60

 pione

-0.59

ozyg

-0.59

æĪ¦

-0.59

RAG

-0.58

roit

-0.58

POSITIVE LOGITS

 Does

1.67

Are

1.48

 Would

1.48

Does

1.43

Did

1.43

Is

1.43

How

1.40

 Could

1.34

Is

1.34

does

1.33

Activations Density 0.622%

titles or phrases that indicate the beginning of an idea or topic

punctuation and question marks

No Comments

No Known Activations