INDEX

Explanations

phrases that express comparative relationships or intensifiers

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

Published

-0.70

igger

-0.62

ULAR

-0.59

dun

-0.57

Explore

-0.56

 eruption

-0.56

bro

-0.54

 concise

-0.54

Begin

-0.53

Sit

-0.53

POSITIVE LOGITS

lihood

0.76

erto

0.68

tainment

0.66

anamo

0.65

pection

0.62

intendent

0.61

 ours

0.61

ernel

0.61

pecting

0.61

ohl

0.60

Activations Density 0.035%

phrases that express comparative relationships or intensifiers

No Comments

No Known Activations