INDEX

Explanations

mentions of specific terms related to technology and names of programming languages

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

WARD

-0.75

ences

-0.70

ITED

-0.69

 downhill

-0.68

INGTON

-0.68

 lockout

-0.67

GMT

-0.66

ENCE

-0.64

ãĤĬ

-0.62

 occup

-0.61

POSITIVE LOGITS

apters

1.14

icago

1.07

ivas

1.06

ieft

0.99

owder

0.99

ocol

0.98

osen

0.98

ampion

0.96

annels

0.96

aos

0.96

Activations Density 3.700%

mentions of specific terms related to technology and names of programming languages

No Comments

No Known Activations