INDEX

Explanations

specific names and terms related to geography or locations

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

 Brill

-0.59

builders

-0.57

 heel

-0.57

 washed

-0.57

 aspirin

-0.56

WARE

-0.56

bridge

-0.56

 flowering

-0.56

 Nordic

-0.56

 Masonic

-0.56

POSITIVE LOGITS

uese

1.13

aru

1.11

oji

1.06

uay

1.06

asu

1.05

rama

1.03

lia

1.02

iy

1.02

icz

1.02

anu

1.02

Activations Density 0.094%

specific names and terms related to geography or locations

No Comments

No Known Activations