INDEX

Explanations

phrases indicating something is not quite as expected or not completely accurate

oai_token-act-pair · gpt-3.5-turbo

phrases indicating uncertainty or qualification

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

orer

-0.67

uana

-0.65

 DRAG

-0.63

ADA

-0.61

kers

-0.60

 instead

-0.58

ifacts

-0.58

recomm

-0.58

 Reviews

-0.58

iop

-0.57

POSITIVE LOGITS

 sure

0.75

ifiable

0.74

orious

0.74

enough

0.73

 Enough

0.73

 enough

0.71

 ready

0.67

ready

0.64

as

0.64

Ready

0.63

Activations Density 0.061%

phrases indicating something is not quite as expected or not completely accurate

phrases indicating uncertainty or qualification

No Comments

No Known Activations