INDEX

Explanations

phrases related to completing, resolving, or decisively addressing a situation or problem

oai_token-act-pair · gpt-3.5-turbo

phrases indicating actions or decisions being made

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

 Digest

-0.75

etheless

-0.67

days

-0.65

 Provides

-0.63

eming

-0.62

 Coverage

-0.61

 Received

-0.58

Ĥİ

-0.56

 Passage

-0.55

lus

-0.55

POSITIVE LOGITS

 aside

1.02

 emphasis

0.89

 brakes

0.84

 jeopardy

0.81

 together

0.80

 pedal

0.79

lid

0.79

 spotlight

0.78

rescent

0.77

 toget

0.77

Activations Density 0.181%

phrases related to completing, resolving, or decisively addressing a situation or problem

phrases indicating actions or decisions being made

No Comments

No Known Activations