INDEX

Explanations

phrases emphasizing efforts to achieve high quality or optimal results

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

kefeller

-0.73

depend

-0.71

usted

-0.68

prus

-0.66

ategory

-0.65

lvl

-0.63

 Canaveral

-0.61

urrencies

-0.60

èĢħ

-0.60

resent

-0.59

POSITIVE LOGITS

 guess

0.77

 endeav

0.73

 imitation

0.72

 impression

0.67

 approximation

0.66

 efforts

0.64

 damned

0.63

seller

0.63

anova

0.63

 intentions

0.62

Activations Density 0.025%

phrases emphasizing efforts to achieve high quality or optimal results

No Comments

No Known Activations