INDEX

Explanations

capitalized terms including "Av" followed by a number

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

hyde

-0.65

ptives

-0.65

 respons

-0.64

 inaccessible

-0.63

FORMATION

-0.62

 Trouble

-0.60

 sense

-0.59

 Barnett

-0.59

 handy

-0.58

utenant

-0.58

POSITIVE LOGITS

ocado

1.23

atars

1.23

ril

1.14

iew

1.07

oided

1.07

ionics

1.05

iol

1.05

ille

1.04

iator

1.03

atar

1.02

Activations Density 0.014%

capitalized terms including "Av" followed by a number

No Comments

No Known Activations