INDEX

Explanations

prepositions and verbs indicating action or intention

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

abiding

-0.92

soDeliveryDate

-0.68

ersed

-0.62

grass

-0.62

 Brilliant

-0.61

pires

-0.61

oplan

-0.60

 Gone

-0.60

 lurking

-0.58

Ĩ

-0.58

POSITIVE LOGITS

 disclose

1.20

 endorse

1.19

 discuss

1.15

 comment

1.14

 specify

1.14

 speculate

1.08

 participate

1.07

 give

1.05

 identify

1.04

 clarify

1.02

Activations Density 0.039%

prepositions and verbs indicating action or intention

No Comments

No Known Activations