INDEX

Explanations

terms related to qualities or characteristics such as flexibility, adaptability, reliability, honesty, accountability, transparency, efficiency, and durability

oai_token-act-pair · gpt-3.5-turbo

concepts related to accountability and integrity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ucha

-0.86

çīĪ

-0.79

nik

-0.76

riber

-0.71

arro

-0.70

han

-0.70

 subscribed

-0.70

JB

-0.69

sov

-0.68

reb

-0.67

POSITIVE LOGITS

 solitude

0.77

 emptiness

0.72

 infiltration

0.71

 heroism

0.70

 complicity

0.68

 detachment

0.67

iveness

0.67

 obscurity

0.67

 responsiveness

0.67

 suppression

0.67

Activations Density 0.789%

terms related to qualities or characteristics such as flexibility, adaptability, reliability, honesty, accountability, transparency, efficiency, and durability

concepts related to accountability and integrity

No Comments

No Known Activations