INDEX

Explanations

terms related to discussions or evaluations by a group of experts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "panel."

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

dx

-0.74

yll

-0.67

 fundament

-0.64

 Pyro

-0.63

 IMAGES

-0.63

DIT

-0.62

 Madison

-0.61

 soluble

-0.60

lihood

-0.59

 Kush

-0.58

POSITIVE LOGITS

ists

1.04

chairs

1.04

 panels

0.94

igans

0.93

 panel

0.93

Panel

0.85

panel

0.85

 convened

0.84

isode

0.82

 chairs

0.81

Activations Density 0.014%

terms related to discussions or evaluations by a group of experts

occurrences of the word "panel."

No Comments

No Known Activations