INDEX

Explanations

sports-related terms, including scores, wins, losses, and player statistics

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

agna

-0.76

icus

-0.64

don

-0.63

eri

-0.60

pain

-0.59

appropriately

-0.58

use

-0.58

advert

-0.58

study

-0.58

ctrl

-0.57

POSITIVE LOGITS

Â½

0.76

ĪĴ

0.75

 underdog

0.74

 aggregate

0.71

 overall

0.71

 overtime

0.70

 halftime

0.70

FG

0.68

TING

0.67

-+-+

0.66

Activations Density 0.530%

sports-related terms, including scores, wins, losses, and player statistics

No Comments

No Known Activations