INDEX

Explanations

mentions of negative aspects or consequences

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

rouse

-0.74

glas

-0.73

ynthesis

-0.72

 Swords

-0.70

orthy

-0.70

aeda

-0.69

aukee

-0.68

 Collider

-0.67

arya

-0.66

ILA

-0.66

POSITIVE LOGITS

der

0.86

 plag

0.80

fully

0.79

 havoc

0.78

ulent

0.78

 inflicted

0.77

 Clown

0.77

asses

0.76

 heap

0.76

ged

0.76

Activations Density 4.465%

mentions of negative aspects or consequences

No Comments

No Known Activations