INDEX

Explanations

key terms related to a specific organization or entity named "MI" and possibly associated with legal or security domains

oai_token-act-pair · gpt-3.5-turbo

mentions of "MI" or related identifiers, likely related to a specific entity or topic

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

holders

-0.83

wagen

-0.82

itatively

-0.78

ional

-0.78

holder

-0.75

 Ragnarok

-0.74

theless

-0.74

ä»£

-0.74

\\\\\\\\

-0.73

comes

-0.72

POSITIVE LOGITS

AMI

1.15

PS

1.08

MI

0.96

HAEL

0.95

KE

0.92

MI

0.87

uppet

0.81

BIL

0.77

PA

0.77

OSH

0.77

Activations Density 0.005%

key terms related to a specific organization or entity named "MI" and possibly associated with legal or security domains

mentions of "MI" or related identifiers, likely related to a specific entity or topic

No Comments

No Known Activations