INDEX

Explanations

phrases related to expressing extremes, particularly in a negative connotation, such as "too much" or "too little."

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

craft

-0.71

inarily

-0.68

ords

-0.67

issance

-0.66

hyde

-0.65

icipated

-0.65

iens

-0.63

idia

-0.61

riots

-0.60

POSITIVE LOGITS

 much

0.84

 busy

0.82

far

0.78

 afraid

0.77

 risky

0.74

oths

0.74

 many

0.74

ls

0.74

 tempting

0.74

 distracting

0.70

Activations Density 0.850%

phrases related to expressing extremes, particularly in a negative connotation, such as "too much" or "too little."

No Comments

No Known Activations