INDEX

Explanations

instances of numbers and quantifiers

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 disposed

-0.68

 condem

-0.68

 redesign

-0.65

 dstg

-0.61

 agre

-0.59

 lapt

-0.57

dayName

-0.56

 piece

-0.55

leep

-0.55

 performance

-0.55

POSITIVE LOGITS

 those

0.83

 these

0.82

us

0.79

those

0.79

them

0.77

 them

0.76

our

0.76

these

0.73

course

0.70

 course

0.67

Activations Density 1.367%

instances of numbers and quantifiers

No Comments

No Known Activations