INDEX

Explanations

comparative phrases indicating increases or decreases in quantity or quality

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

abama

-0.73

{}

-0.69

ode

-0.67

cture

-0.66

gee

-0.64

atan

-0.63

ouse

-0.62

assis

-0.62

era

-0.61

iste

-0.61

POSITIVE LOGITS

 importantly

1.43

 emphasis

0.91

 reliance

0.90

 likely

0.88

 reliant

0.83

 inclined

0.82

natureconservancy

0.75

 prone

0.75

 than

0.75

 sophisticated

0.75

Activations Density 0.022%

comparative phrases indicating increases or decreases in quantity or quality

No Comments

No Known Activations