INDEX

Explanations

comparisons related to ease or difficulty

oai_token-act-pair · gpt-3.5-turbo

the word "easier" 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 @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Saint

-0.74

inals

-0.72

Origin

-0.70

 Pall

-0.69

wine

-0.68

eters

-0.66

reen

-0.66

hips

-0.66

alian

-0.65

 Nether

-0.64

POSITIVE LOGITS

 than

1.03

 forgiving

0.96

 prey

0.95

 compr

0.88

 "$:/

0.82

 manageable

0.76

 readable

0.73

 easy

0.72

iquid

0.71

 behaved

0.68

Activations Density 0.012%

comparisons related to ease or difficulty

the word "easier" in various contexts

No Comments

No Known Activations