INDEX

Explanations

phrases indicating a purpose or intention with the word "so" followed by an action or consequence

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

wheel

-0.62

 Tent

-0.61

 Visit

-0.59

 understatement

-0.58

been

-0.57

major

-0.56

 hijab

-0.55

ectomy

-0.55

 Element

-0.55

 Verse

-0.55

POSITIVE LOGITS

 maximize

1.06

 minimize

0.92

 insure

0.92

 avoid

0.92

avoid

0.90

 facilitate

0.89

 conserve

0.87

 maxim

0.87

 optimize

0.84

 ensure

0.82

Activations Density 1.735%

phrases indicating a purpose or intention with the word "so" followed by an action or consequence

No Comments

No Known Activations