INDEX

Explanations

phrases indicating exclusivity or uniqueness

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

env

-0.66

insula

-0.64

ahime

-0.64

idon

-0.59

ahead

-0.58

 PLUS

-0.57

comb

-0.55

 NAACP

-0.53

duino

-0.53

align

-0.52

POSITIVE LOGITS

 marginally

1.01

ices

0.87

ICES

0.84

 insofar

0.81

 spor

0.80

 incidentally

0.75

 mildly

0.72

 scratches

0.71

 kidding

0.71

onse

0.68

Activations Density 0.069%

phrases indicating exclusivity or uniqueness

No Comments

No Known Activations