INDEX

Explanations

the word "Wat" in various contexts, suggesting a focus on the concept of observation or watchfulness

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

Reviewer

-0.81

pmwiki

-0.78

wcs

-0.66

interstitial

-0.65

staking

-0.64

taboola

-0.61

 Haas

-0.61

 appropriation

-0.59

 prevailing

-0.59

Enlarge

-0.59

POSITIVE LOGITS

terson

1.52

cher

1.14

chers

1.09

chens

1.05

ches

1.04

ched

1.03

chen

1.02

combe

0.97

chet

0.96

erv

0.91

Activations Density 0.004%

the word "Wat" in various contexts, suggesting a focus on the concept of observation or watchfulness

No Comments

No Known Activations