INDEX

Explanations

phrases related to inappropriate behavior

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

IFrame

Link

Negative Logits

ership

-0.87

rage

-0.81

utra

-0.79

elsen

-0.78

ingham

-0.78

ynthesis

-0.77

fighters

-0.76

stones

-0.76

raq

-0.76

itar

-0.75

POSITIVE LOGITS

 touching

0.91

 behavior

0.90

 Parenthood

0.87

ities

0.87

 behaviour

0.85

 enrichment

0.84

 behaviors

0.80

 interference

0.78

 interfering

0.76

 behaviours

0.74

Activations Density 0.032%

phrases related to inappropriate behavior

No Known Activations