INDEX

Explanations

locations within a building related to events or conflicts

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 Rounds

-0.72

nir

-0.72

utan

-0.72

CHAT

-0.68

ï¸ı

-0.66

ãģĦ

-0.65

 Marketable

-0.62

Uz

-0.60

yon

-0.60

OTH

-0.59

POSITIVE LOGITS

 doors

1.07

 entrances

1.06

 floor

1.00

 entrance

0.99

room

0.98

house

0.97

 door

0.95

itory

0.94

acle

0.94

keeper

0.93

Activations Density 4.812%

locations within a building related to events or conflicts

No Comments

No Known Activations