INDEX

Explanations

the word "it"

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

icons

-0.66

avage

-0.66

hips

-0.65

collection

-0.60

represented

-0.57

VA

-0.57

DATA

-0.57

cence

-0.57

iling

-0.57

animal

-0.57

POSITIVE LOGITS

alian

1.04

chy

0.94

 raining

0.92

iner

0.88

unes

0.84

self

0.81

asca

0.80

 horizont

0.78

 Anyway

0.74

MpServer

0.72

Activations Density 0.110%

the word "it"

No Comments

No Known Activations