INDEX

Explanations

phrases related to lists or collections of items or topics

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

matter

-0.61

hya

-0.59

 disposed

-0.59

 condem

-0.58

iture

-0.57

leep

-0.57

 lapt

-0.56

 buffet

-0.55

 shade

-0.55

zeb

-0.55

POSITIVE LOGITS

 those

0.70

few

0.67

those

0.67

arching

0.66

icial

0.65

arb

0.65

 Decay

0.63

 many

0.61

 them

0.61

few

0.60

Activations Density 0.352%

phrases related to lists or collections of items or topics

No Comments

No Known Activations