INDEX

Explanations

phrases related to advocating for specific actions or beliefs

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

ciation

-0.71

 doubtless

-0.70

PG

-0.64

inally

-0.64

eatures

-0.63

zzi

-0.62

 nifty

-0.60

albeit

-0.60

ionics

-0.60

ector

-0.59

POSITIVE LOGITS

 anymore

1.27

nor

1.02

any

0.95

 unnecessarily

0.92

 anything

0.88

 blindly

0.88

 rash

0.86

 frivol

0.86

 jeopard

0.85

 disrespect

0.85

Activations Density 0.416%

phrases related to advocating for specific actions or beliefs

No Comments

No Known Activations