INDEX

Explanations

references to qualities that may depict criticism or commentary

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

rompt

-0.77

heid

-0.70

icer

-0.65

pload

-0.62

alid

-0.59

utive

-0.59

thora

-0.59

 Mellon

-0.59

asar

-0.58

moil

-0.58

POSITIVE LOGITS

 importantly

1.33

afa

1.06

 notably

0.94

likely

0.92

 likely

0.90

 important

0.83

 prominently

0.83

 egreg

0.83

 notable

0.75

 mornings

0.75

Activations Density 0.665%

references to qualities that may depict criticism or commentary

No Comments

No Known Activations