INDEX

Explanations

occurrences of the letter 's'

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

 trusts

-0.71

 subcontract

-0.64

 guarding

-0.64

 penetrated

-0.64

 attaches

-0.64

 exercised

-0.62

 sacrific

-0.62

 intimidated

-0.61

 graz

-0.60

eus

-0.60

POSITIVE LOGITS

 anniversary

0.94

ELF

0.86

 edition

0.86

ixty

0.83

herry

0.81

ullivan

0.80

omal

0.79

 inaugural

0.79

querade

0.79

outh

0.77

Activations Density 0.055%

occurrences of the letter 's'

No Comments

No Known Activations