INDEX

Explanations

programming-related terminology, especially related to fetching, parsing, compiling, and handling errors

oai_token-act-pair · gpt-3.5-turbo

technical terms and phrases related to programming, specifically involving error handling and data processing

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

amiya

-0.85

audi

-0.83

clerosis

-0.79

arak

-0.76

anship

-0.73

soDeliveryDate

-0.72

staff

-0.71

raud

-0.71

trade

-0.70

ayne

-0.70

POSITIVE LOGITS

 cached

1.27

 dynamically

1.08

 cache

1.02

 output

1.00

 JSON

0.99

UTF

0.99

 specified

0.96

XML

0.95

 HTTP

0.94

 automatically

0.94

Activations Density 0.424%

programming-related terminology, especially related to fetching, parsing, compiling, and handling errors

technical terms and phrases related to programming, specifically involving error handling and data processing

No Comments

No Known Activations