INDEX

Explanations

unnamed formatting tokens

oai_token-act-pair · gpt-3.5-turbo

numerical values or statistics in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 skelet

-0.84

 wardrobe

-0.73

 farewell

-0.73

 knockout

-0.71

 courtroom

-0.70

 vict

-0.69

 mint

-0.69

 circus

-0.68

 roundup

-0.68

 parting

-0.67

POSITIVE LOGITS

However

1.59

He

1.59

SPONSORED

1.55

Refer

1.52

Asked

1.51

"(

1.51

Such

1.50

Others

1.49

"[

1.44

Besides

1.43

Activations Density 0.324%

unnamed formatting tokens

numerical values or statistics in various contexts

No Comments

No Known Activations