INDEX

Explanations

infinitive verbs that indicate intentions or threats

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

Reviewer

-0.82

 Parables

-0.77

abiding

-0.74

emis

-0.71

erning

-0.71

iencies

-0.71

ECO

-0.70

quickShipAvailable

-0.70

cellent

-0.70

ortunate

-0.69

POSITIVE LOGITS

 blackmail

1.03

 arrest

1.01

 wrath

0.98

sue

0.98

 veto

0.97

 boycott

0.97

 drown

0.95

 derail

0.93

 retribution

0.92

 sabotage

0.92

Activations Density 0.069%

infinitive verbs that indicate intentions or threats

No Comments

No Known Activations