INDEX

Explanations

email addresses and technical terms related to digital information and processing

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

CLR

-0.69

DERR

-0.65

this

-0.59

\<

-0.57

 Doors

-0.55

 Names

-0.55

hern

-0.54

 Eyes

-0.54

Unknown

-0.54

 trucks

-0.54

POSITIVE LOGITS

cture

0.90

 belongs

0.77

 alone

0.76

illet

0.71

stros

0.71

isphere

0.71

OULD

0.70

ranch

0.67

rame

0.67

 represents

0.66

Activations Density 1.363%

email addresses and technical terms related to digital information and processing

No Comments

No Known Activations