INDEX

Explanations

words related to underground or hidden activities, often associated with danger or mystery

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Brotherhood

-0.62

 embodiments

-0.62

 embr

-0.61

 prisoners

-0.61

 chloride

-0.60

CTV

-0.59

 Comfort

-0.59

CVE

-0.59

grave

-0.58

 inmates

-0.58

POSITIVE LOGITS

yright

1.00

ops

0.97

olitan

0.92

odcast

0.86

edia

0.84

imus

0.82

rotein

0.81

iate

0.79

olicy

0.79

inion

0.78

Activations Density 0.036%

words related to underground or hidden activities, often associated with danger or mystery

No Comments

No Known Activations