INDEX

Explanations

instances of the word "have."

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

ilation

-0.71

stre

-0.70

FIG

-0.70

soon

-0.65

 Slowly

-0.64

attery

-0.63

Introdu

-0.61

Soc

-0.61

odge

-0.61

 Initially

-0.60

POSITIVE LOGITS

 recourse

0.96

 enough

0.93

 permission

0.93

 anymore

0.91

 backups

0.88

 sufficient

0.88

 adequate

0.85

 access

0.85

stood

0.85

any

0.84

Activations Density 0.044%

instances of the word "have."

No Comments

No Known Activations