INDEX

Explanations

body parts and related actions

oai_token-act-pair · gpt-3.5-turbo

body part references

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

ãĤ¤ãĥĪ

-0.70

Ń·

-0.67

Vers

-0.67

 Virt

-0.66

 Sell

-0.65

afort

-0.62

Veh

-0.62

Sit

-0.62

 Lank

-0.60

Neg

-0.60

POSITIVE LOGITS

pan

0.83

ache

0.80

hoe

0.79

 cavity

0.78

 cheeks

0.75

 recorder

0.73

 footprint

0.73

piece

0.73

 paws

0.71

 muscles

0.71

Activations Density 0.128%

body parts and related actions

body part references

No Comments

No Known Activations