INDEX

Explanations

words related to specific character names and titles

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Downloadha

-0.73

reddits

-0.62

yip

-0.57

 princip

-0.57

ģ«

-0.56

 mosqu

-0.56

osate

-0.56

 NETWORK

-0.56

 toget

-0.55

 maker

-0.54

POSITIVE LOGITS

ophon

0.74

ouston

0.72

theless

0.68

entials

0.66

AFB

0.66

anced

0.65

bg

0.62

monds

0.62

andestine

0.61

zona

0.61

Activations Density 0.158%

words related to specific character names and titles

No Comments

No Known Activations