INDEX

Explanations

word tokens related to storage, such as storage box, SD card, and memory

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

kok

-0.77

ulhu

-0.75

ĪĴ

-0.72

hua

-0.72

enburg

-0.71

enberg

-0.70

 Cosponsors

-0.70

zai

-0.68

pheus

-0.68

atown

-0.67

POSITIVE LOGITS

 storage

1.15

 storing

1.01

storage

1.00

 capacity

0.97

 shelves

0.97

 Capacity

0.96

capacity

0.95

 dumps

0.93

 depot

0.93

 capacities

0.86

Activations Density 1.116%

word tokens related to storage, such as storage box, SD card, and memory

No Comments

No Known Activations