INDEX

Explanations

statements evaluating the goodness or advisability of various actions or ideas

oai_token-act-pair · gpt-3.5-turbo

qualitative evaluations or judgments about what is considered "good."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

RAW

-0.79

doms

-0.77

 Rooms

-0.73

 actionGroup

-0.72

lees

-0.71

hm

-0.70

ngth

-0.69

ĸļ

-0.69

ATURES

-0.68

andestine

-0.67

POSITIVE LOGITS

 idea

1.42

 thing

1.37

 predictor

1.11

 excuse

1.11

 indicator

1.10

 indication

1.07

fit

1.06

 outcome

1.04

 place

1.04

 approximation

1.00

Activations Density 0.068%

statements evaluating the goodness or advisability of various actions or ideas

qualitative evaluations or judgments about what is considered "good."

No Comments

No Known Activations