INDEX

Explanations

punctuation marks

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

 spont

-0.72

 drawer

-0.66

gag

-0.65

 vanity

-0.62

 rigs

-0.61

 appe

-0.60

wagen

-0.59

peg

-0.59

 everyday

-0.59

 crude

-0.58

POSITIVE LOGITS

actionDate

0.95

][

0.84

ĸļ

0.77

taboola

0.76

à¼

0.76

align

0.76

::::::::

0.73

ojure

0.72

mosp

0.70

lement

0.70

Activations Density 0.782%

punctuation marks

No Comments

No Known Activations