INDEX

Explanations

characters or persons of interest, particularly actors

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

Not in Any Lists

No Comments

Negative Logits

ornings

-0.66

road

-0.64

awks

-0.63

 Recipes

-0.61

hare

-0.61

ourcing

-0.60

lap

-0.60

poons

-0.58

wcsstore

-0.58

creen

-0.58

POSITIVE LOGITS

's

0.91

 whom

0.89

 personally

0.82

 himself

0.80

$.

0.74

owicz

0.72

who

0.71

rer

0.71

 deceased

0.70

liest

0.69

Activations Density 0.286%

characters or persons of interest, particularly actors

No Comments

No Known Activations