INDEX

Explanations

terms related to varying degrees of severity or leniency in policies and approaches

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

 untold

-0.77

ividual

-0.73

Haunted

-0.65

Missing

-0.64

iverse

-0.64

 Halls

-0.64

library

-0.63

ousands

-0.63

ames

-0.63

 Nightmares

-0.63

POSITIVE LOGITS

 stance

1.36

 stances

1.14

 diplomacy

1.12

 tactics

1.10

 demeanor

1.10

 posture

1.10

 attitude

1.10

 rhetoric

1.06

 negotiators

1.00

 diplomatic

0.96

Activations Density 0.117%

terms related to varying degrees of severity or leniency in policies and approaches

No Comments

No Known Activations