INDEX

Explanations

phrases that emphasize importance or make comparisons

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_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

hid

-0.67

bridge

-0.65

hl

-0.65

retched

-0.65

sil

-0.65

stood

-0.65

ences

-0.64

dl

-0.64

file

-0.63

orc

-0.63

POSITIVE LOGITS

 considering

1.06

 since

0.83

 insofar

0.80

 given

0.79

given

0.75

 compared

0.72

 geared

0.69

 egregious

0.68

 because

0.67

 those

0.66

Activations Density 0.029%

phrases that emphasize importance or make comparisons

No Comments

No Known Activations