INDEX

Explanations

phrases related to critiques or negative commentary

oai_token-act-pair · gpt-3.5-turbo

references to quantity or amounts

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

ĸļ

-0.84

 agre

-0.82

 condem

-0.70

dayName

-0.69

oler

-0.68

 depot

-0.67

 disposed

-0.65

acus

-0.64

posal

-0.63

 ende

-0.62

POSITIVE LOGITS

arching

0.83

things

0.64

us

0.63

icial

0.63

legged

0.62

 unlucky

0.62

tan

0.61

 whom

0.61

 lesser

0.58

outed

0.57

Activations Density 0.104%

phrases related to critiques or negative commentary

references to quantity or amounts

No Comments

No Known Activations