INDEX

Explanations

words related to conflicting or opposing information

oai_token-act-pair · gpt-3.5-turbo

concepts and references related to contradictions and conflicting ideas

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.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

ids

-0.66

gain

-0.66

nab

-0.66

Geh

-0.64

tu

-0.63

 Nets

-0.62

ju

-0.62

agar

-0.61

hound

-0.61

 Deep

-0.61

POSITIVE LOGITS

 contradictory

3.22

 conflicting

2.79

 contradictions

2.76

 contradict

2.73

 contradiction

2.71

 contradicted

2.33

 contradicts

2.20

 inconsistent

2.19

 inconsistency

2.13

 inconsistencies

2.00

Activations Density 0.030%

words related to conflicting or opposing information

concepts and references related to contradictions and conflicting ideas

No Comments

No Known Activations