INDEX

Explanations

words related to initiating or stopping actions, often in a figurative or metaphorical sense

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

lav

-0.69

byter

-0.68

andra

-0.66

ollen

-0.66

solid

-0.64

atha

-0.62

Spread

-0.61

iannopoulos

-0.61

haps

-0.60

heid

-0.60

POSITIVE LOGITS

 prematurely

0.91

washer

0.81

 brakes

0.76

 sooner

0.73

mination

0.69

doors

0.69

 offensively

0.66

 timer

0.65

 notch

0.64

 salute

0.64

Activations Density 0.043%

words related to initiating or stopping actions, often in a figurative or metaphorical sense

No Comments

No Known Activations