INDEX

Explanations

phrases indicating contrast or negation

oai_token-act-pair · gpt-3.5-turbo

references to significant events or facts that are often exaggerated or misrepresented

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

WT

-0.73

acca

-0.67

arine

-0.66

igate

-0.66

agos

-0.63

ecast

-0.61

inas

-0.60

 Travels

-0.60

 Dialogue

-0.58

ukong

-0.58

POSITIVE LOGITS

 nonetheless

1.57

 nevertheless

1.35

etheless

1.02

 still

0.82

 retained

0.75

 darn

0.74

 strangely

0.73

 awfully

0.72

 proble

0.71

 undeniably

0.70

Activations Density 1.184%

phrases indicating contrast or negation

references to significant events or facts that are often exaggerated or misrepresented

No Comments

No Known Activations