INDEX

Explanations

negations or phrases indicating conditionality or exceptions

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

Topic

-0.71

\\\\\\\\

-0.66

Medium

-0.65

cx

-0.64

ãĤµ

-0.64

Values

-0.63

Specific

-0.62

ãģ®ç

-0.62

onic

-0.62

Course

-0.61

POSITIVE LOGITS

 existed

0.83

 intervened

0.81

 happened

0.75

hin

0.73

terday

0.73

 bailed

0.64

 complications

0.63

bucks

0.62

essors

0.62

hai

0.60

Activations Density 0.086%

negations or phrases indicating conditionality or exceptions

No Comments

No Known Activations