INDEX

Explanations

statements that introduce additional information or context in the text

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

crow

-0.73

 throats

-0.69

ocene

-0.68

abies

-0.64

 hearts

-0.62

RED

-0.60

fell

-0.58

 Balkans

-0.58

ICAN

-0.57

stay

-0.57

POSITIVE LOGITS

 importantly

0.91

entimes

0.83

,.

0.79

.,

0.77

 guiActiveUn

0.71

è£ħ

0.71

 interestingly

0.70

adays

0.67

ãĥ»

0.66

Activations Density 0.014%

statements that introduce additional information or context in the text

No Comments

No Known Activations