INDEX

Explanations

the word 'thing'

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

DOS

-0.76

zeb

-0.74

inventoryQuantity

-0.71

ãĥ¼ãĥ

-0.71

vez

-0.71

ahime

-0.70

zek

-0.68

claimed

-0.67

taboola

-0.67

ãĥ«

-0.65

POSITIVE LOGITS

rolling

0.65

iverse

0.63

soever

0.63

 else

0.62

 Constantine

0.62

icularly

0.60

 Carm

0.59

 altogether

0.59

 entirely

0.57

 apiece

0.57

Activations Density 0.033%

the word 'thing'

No Comments

No Known Activations