INDEX

Explanations

terms related to "unknown" or "undeveloped" concepts

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

ership

-0.75

 Catal

-0.70

tsky

-0.70

 Tycoon

-0.70

 understatement

-0.69

 Gazette

-0.68

Reviewer

-0.67

 Duchess

-0.63

 Charl

-0.63

 rejection

-0.62

POSITIVE LOGITS

ored

1.16

oded

1.15

enced

1.06

itable

1.04

structed

1.03

ired

1.02

overed

1.01

ought

1.01

velop

0.99

served

0.96

Activations Density 0.011%

terms related to "unknown" or "undeveloped" concepts

No Comments

No Known Activations