INDEX

Explanations

references to communication and written expression in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

SPA

-0.69

icio

-0.64

opolis

-0.62

HAS

-0.60

htar

-0.59

Virgin

-0.57

Lu

-0.56

leneck

-0.55

Bir

-0.55

panic

-0.55

POSITIVE LOGITS

 themselves

0.98

 respectively

0.92

 collectively

0.86

 individually

0.83

 apiece

0.80

geries

0.77

 ones

0.77

 expire

0.75

 converge

0.75

 clustered

0.74

Activations Density 1.033%

references to communication and written expression in different contexts

No Comments

No Known Activations