INDEX

Explanations

phrases related to instructions or specific actions

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

hap

-0.82

uga

-0.79

iosyn

-0.77

OST

-0.74

uay

-0.71

ordable

-0.70

regular

-0.69

otos

-0.69

ulz

-0.68

obal

-0.68

POSITIVE LOGITS

 ratio

1.24

 syndrome

1.15

 Ratio

1.07

 ratios

1.03

 clause

0.99

 trope

0.95

 mentality

0.95

 fallacy

0.94

 Syndrome

0.90

 initiative

0.90

Activations Density 1.823%

phrases related to instructions or specific actions

No Comments

No Known Activations