INDEX

Explanations

phrases indicating causation or reasoning, particularly with the word "because."

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

lez

-0.72

wn

-0.71

robe

-0.69

agin

-0.69

ax

-0.69

mint

-0.66

lem

-0.65

nin

-0.64

Gas

-0.64

yan

-0.63

POSITIVE LOGITS

 they

1.04

 nobody

0.90

 there

0.89

it

0.84

 unlike

0.82

 otherwise

0.81

we

0.79

 THEY

0.79

*/(

0.78

he

0.75

Activations Density 0.561%

phrases indicating causation or reasoning, particularly with the word "because."

No Comments

No Known Activations