INDEX

Explanations

occurrences of the word "entered" and related actions involving entrance or access

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

torch.float32

Hook Point

blocks.8.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

 pros

-0.75

 tongue

-0.72

accompan

-0.71

IPP

-0.69

 elbows

-0.68

 endorsements

-0.66

eday

-0.66

 juggling

-0.65

azeera

-0.64

umbnails

-0.63

POSITIVE LOGITS

 unnoticed

0.84

 breach

0.80

tained

0.79

 Paddock

0.78

 undet

0.77

 doors

0.77

 rooms

0.75

bery

0.75

 bedrooms

0.75

 breached

0.73

Activations Density 0.046%

occurrences of the word "entered" and related actions involving entrance or access

No Comments

No Known Activations