INDEX

Explanations

abbreviations or terms commonly used in discussions of network protocols and data formats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

gie

-0.84

hower

-0.79

hold

-0.76

hips

-0.74

neau

-0.70

avia

-0.69

reau

-0.68

flies

-0.68

abeth

-0.68

ween

-0.67

POSITIVE LOGITS

olicy

1.08

ublic

1.08

odcast

1.02

arser

0.98

NP

0.97

ivot

0.96

ixel

0.92

QL

0.92

IX

0.91

ACE

0.91

Activations Density 0.007%

abbreviations or terms commonly used in discussions of network protocols and data formats

No Comments

No Known Activations