INDEX

Explanations

comparisons indicating improvement or preference

oai_token-act-pair · gpt-3.5-turbo

comparative terms indicating preference or judgment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

utm

-0.75

eur

-0.70

ulating

-0.65

ategory

-0.62

uly

-0.62

 Greenpeace

-0.61

urer

-0.61

inous

-0.61

itory

-0.61

naires

-0.60

POSITIVE LOGITS

yet

0.99

 than

0.93

 Than

0.88

than

0.84

ment

0.81

 still

0.79

 behaved

0.75

 luck

0.73

 safe

0.71

bye

0.71

Activations Density 0.033%

comparisons indicating improvement or preference

comparative terms indicating preference or judgment

No Comments

No Known Activations