INDEX

Explanations

numerical values associated with military specifications or classifications

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

ktop

-0.69

estamp

-0.69

practice

-0.66

elist

-0.65

ools

-0.65

govtrack

-0.63

weekly

-0.63

âĸ¬

-0.62

healthy

-0.62

falls

-0.61

POSITIVE LOGITS

bis

0.93

Bs

0.90

SO

0.83

 Abrams

0.79

 dstg

0.77

CV

0.76

IOR

0.75

 Biological

0.75

BN

0.75

 Bomber

0.75

Activations Density 0.031%

numerical values associated with military specifications or classifications

No Comments

No Known Activations