INDEX

Explanations

terms related to control and commands

oai_token-act-pair · gpt-3.5-turbo

terms related to control mechanisms or functionalities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

tein

-0.92

olit

-0.73

iatus

-0.72

iversary

-0.72

forth

-0.69

tern

-0.68

ruary

-0.67

entimes

-0.66

ead

-0.65

 Chrys

-0.65

POSITIVE LOGITS

 levers

0.93

 steering

0.86

eering

0.84

 impulses

0.84

 controls

0.80

 controller

0.79

 knob

0.79

ulators

0.79

 controllers

0.79

 pedal

0.78

Activations Density 0.030%

terms related to control and commands

terms related to control mechanisms or functionalities

No Comments

No Known Activations