INDEX

Explanations

punctuation marks, such as commas

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

 drawer

-0.70

 spont

-0.69

gag

-0.66

 rigs

-0.62

 misunder

-0.60

ussy

-0.58

 appe

-0.57

 vanity

-0.57

 everyday

-0.57

 ordinary

-0.57

POSITIVE LOGITS

actionDate

0.95

][

0.79

align

0.71

ĸļ

0.71

ojure

0.70

taboola

0.70

à¼

0.70

then

0.69

,,,,,,,,

0.66

::::::::

0.65

Activations Density 0.278%

punctuation marks, such as commas

No Comments

No Known Activations