rotta_rs 0.0.6

a Deep Learning library with rust language
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451

## 📦 Version: `0.0.6`

### ✨ New Features
- `lstm function`
```rust
fn main() {
    let mut model = Module::init();
    let lstm = model.lstm_init(6);

    let tensor = Tensor::arange(0..6)
        .to_shape(vec![1, 6])
        .collect();
    println!("{}", tensor);
    // [
    //  [0.0, 1.0, 2.0, 3.0, 4.0, 5.0]
    // ]

    let cell_hidden = lstm.forward(&tensor, None); //(&Tensor, Option<LSTMCellHidden>)
    println!("cell:\n{}", cell_hidden.cell);
    // cell:
    // [
    //  [-0.35452980609595336, -0.5292895572406782, 0.7244741335616286, -0.4735260362646916, -0.9515427099660342, -0.2374937433097324]
    // ]

    println!("hidden:\n{}", cell_hidden.hidden)
    // hidden:
    // [
    //  [-0.23028978337822023, -0.0435225356725021, 0.36147911006385536, -0.09738843166346652, -0.705630337819917, -0.051854840060327305]
    // ]
}
```

```rust
fn main() {
    LSTMCellHidden {
        cell: Tensor::new([[0.0]]),
        hidden: Tensor::new([[0.0]]),
    };
}
```

- `gru function`
```rust
fn main() {
    let mut model = Module::init();
    let lstm = model.gru_init(6);

    let tensor = Tensor::arange(0..6)
        .to_shape(vec![1, 6])
        .collect();
    println!("{}", tensor);
    // [
    //  [0.0, 1.0, 2.0, 3.0, 4.0, 5.0]
    // ]


    let hidden = lstm.forward(&tensor, None); // (&Tensor, Option<Tensor>)
    println!("hidden:\n{}", hidden);
    // hidden:
    // [
    //  [0.2838769389632212, 0.30500443468104915, 0.03277033032455531, -0.4399917698902819, -0.9237180542375861, 0.3168019973639146]
    // ]
}
```

- `argmax method`
```rust
fn main() {
    let tensor = Tensor::rand(vec![3, 4]);
    println!("{}", tensor);
    // [
    //  [0.490653500989335, 0.9084708072918432, 0.5516854111601106, 0.38535295859939467]
    //  [0.5510768003422782, 0.04506790076914613, 0.36501508148354644, 0.48345186596013223]
    //  [0.05192318702409915, 0.03138051383420948, 0.5086337769325273, 0.021017655640771404]
    // ]

    let max = tensor.argmax(0);
    println!("{}", max);
    // [1.0, 0.0, 0.0, 1.0]

    let max = tensor.argmax(1);
    println!("{}", max)
    // [1.0, 0.0, 2.0]
}
```

- `argmin method`
```rust
fn main() {
    let tensor = Tensor::rand(vec![3, 4]);
    println!("{}", tensor);
    // [
    //  [0.8611219295171262, 0.9410276855253755, 0.9133598099213944, 0.22062707185602048]
    //  [0.47193516661684776, 0.7818072906711374, 0.8048492003479746, 0.9925399063075784]
    //  [0.3559389967244023, 0.3472829046036767, 0.7791381493184755, 0.8910867638091713]
    // ]

    let max = tensor.argmin(0);
    println!("{}", max);
    // [2.0, 2.0, 2.0, 0.0]

    let max = tensor.argmin(1);
    println!("{}", max)
    // [3.0, 0.0, 1.0]
}
```

- update `arange` method
```rust
fn main() {
    let tensor = Tensor::arange(0..12).collect();
    println!("{}", tensor);
    //[0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0]


    let tensor = Tensor::arange(0..24)
        .step(2)
        .to_shape(vec![3, 4])
        .map(|x| x * 2.0)
        .collect();
    println!("{}", tensor)
    // [
    //  [0.0, 4.0, 8.0, 12.0]
    //  [16.0, 20.0, 24.0, 28.0]
    //  [32.0, 36.0, 40.0, 44.0]
    // ]
}
```

- update `concat` method
```rust
fn main() {
    let tensor_a = Tensor::arange(0..3)
        .to_shape(vec![1, 3])
        .collect();
    let tensor_b = Tensor::arange(3..6)
        .to_shape(vec![1, 3])
        .collect();

    // before 0.0.6
    let tensors = vec![&tensor_a, &tensor_b];
    let concat = concat(tensors, 0);
    println!("{}", concat);
    // [
    //  [0.0, 1.0, 2.0]
    //  [3.0, 4.0, 5.0]
    // ]

    // after 0.0.6
    let concat = vec![&tensor_a, &tensor_b].concat_tensor(0);
    println!("{}", concat);
    // [
    //  [0.0, 1.0, 2.0]
    //  [3.0, 4.0, 5.0]
    // ]

    let concat = vec![tensor_a, tensor_b].concat_tensor(0);
    println!("{}", concat)
    // [
    //  [0.0, 1.0, 2.0]
    //  [3.0, 4.0, 5.0]
    // ]
}
```


- `flatten method`
```rust
fn main() {
    let tensor = Tensor::arange(0..12)
        .to_shape(vec![3, 4])
        .collect();
    println!("{}", tensor);
    // [
    //  [0.0, 1.0, 2.0, 3.0]
    //  [4.0, 5.0, 6.0, 7.0]
    //  [8.0, 9.0, 10.0, 11.0]
    // ]

    let flat = tensor.flatten();
    println!("{}", flat);
    // [0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0]
}
```

- `sin, cos, tan method`
```rust
fn main() {
    let tensor = Tensor::arange(0..12)
        .to_shape(vec![3, 4])
        .collect();
    println!("{}", tensor);
    // [
    //  [0.0, 1.0, 2.0, 3.0]
    //  [4.0, 5.0, 6.0, 7.0]
    //  [8.0, 9.0, 10.0, 11.0]
    // ]


    let sin = tensor.sin();

    let cos = tensor.cos();

    let tan = tensor.tan();
}
```

- update `embedding` function
```rust
fn main() {
    let mut model = Module::init();
    let embedding = model.embedding_init(6, 4); //(vocab_num: usize, hidden: usize)
    println!("{}", embedding.parameter);
    // [
    //  [0.1818961923066713, 0.450275407672484, -0.0724835971434803, 0.12736052119734032]
    //  [-0.21140612085881738, -0.35004112970967505, -0.19195944040209034, 0.3038727671756267]
    //  [0.27124878080285697, -0.2614147356186607, 0.006866870340068942, 0.4018031720487738]
    //  [0.45699026891614114, 0.09297857032433399, 0.22879848650077528, -0.3208635907368216]
    //  [0.07539314660070717, -0.32823770909937444, -0.055947665079024045, -0.42360899591141266]
    //  [-0.34544166573751034, -0.17956923353240994, 0.1996879885620404, -0.42159451133374315]
    // ]

    // using the float data type because the usize data type that is generally used in indexing is not yet available
    let token = Tensor::new([0.0, 2.0, 5.0]);
    let embedded = embedding.forward(&token);
    println!("{}", embedded);
    // [
    //  [0.1818961923066713, 0.450275407672484, -0.0724835971434803, 0.12736052119734032]
    //  [0.27124878080285697, -0.2614147356186607, 0.006866870340068942, 0.4018031720487738]
    //  [-0.34544166573751034, -0.17956923353240994, 0.1996879885620404, -0.42159451133374315]
    // ]
}
```


### 🚀 Optimizations
- Optimizing Slicing Method

### 🔧 Change
- replace `StdRng` to `chacha8rng`

# 📦 Version: `0.0.5`

### ✨ New Features
- `arange` method for create tensor
```rust
    let tensor = Tensor::arange(0, 10, 2); // (start, stop, step)
    println!("{}", tensor) // [0.0, 2.0, 4.0, 6.0, 8.0]
```

- `concat method for combining tensors in vectors`
```rust
fn main() {
    let tensor_a = Tensor::new([[1.0, 2.0, 3.0, 4.0, 5.0]]);
    let tensor_b = Tensor::new([[6.0, 7.0, 8.0, 9.0, 10.0]]);
    let vector = vec![&tensor_a, &tensor_b];

    let tensor = concat(vector, 0);
    println!("{}", tensor);     // [
                                //  [1.0, 2.0, 3.0, 4.0, 5.0]
                                //  [6.0, 7.0, 8.0, 9.0, 10.0]
                                // ]    
}
```

- `new method for slicing`
```rust
fn main() {
    let tensor_a = Tensor::arange(0, 12, 1).reshape(vec![-1, 3]);
    println!("{}", tensor_a);
    // [
    //  [0.0, 1.0, 2.0]
    //  [3.0, 4.0, 5.0]
    //  [6.0, 7.0, 8.0]
    //  [9.0, 10.0, 11.0]
    // ]

    // before 0.0.5
    let slicing = tensor_a.slice(vec![ArrSlice(Some(0), Some(2)), ArrSlice(Some(1), None)]);
    println!("{}", slicing);
    // [
    //  [1.0, 2.0]
    //  [4.0, 5.0]
    // ]

    // 0.0.5
    let slicing = tensor_a.slice(vec![r(0..2), r(1..)]);
    println!("{}", slicing)
    // [
    //  [1.0, 2.0]
    //  [4.0, 5.0]
    // ]
}
```

- `new method for sum axis and sum axis keep dim`
```rust
fn main() {
    let tensor_a = Tensor::arange(0, 12, 1).reshape(vec![-1, 3]);
    println!("{}", tensor_a);
    // [
    //  [0.0, 1.0, 2.0]
    //  [3.0, 4.0, 5.0]
    //  [6.0, 7.0, 8.0]
    //  [9.0, 10.0, 11.0]
    // ]

    // before 0.0.5
    // let sum = tensor_a.sum_axis(0); // in version 0.0.5, it can no longer be done

    // 0.0.5
    let slicing = tensor_a.sum_axis(&[0]);
    println!("{}", slicing);
    // [18.0, 22.0, 26.0]

    let slicing = tensor_a.sum_axis_keep_dim(&[0, 1]);
    println!("{}", slicing);
    // [
    //  [66.0]
    // ]
}
```

- `mean` & `mean axis` & `mean axis keep dim`

see more details in guide.md [tensor section](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/1_tensor.md)

- `RMSProp`

see more details in guide.md [Optimazer section](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/5_Optimazer.md)

- `Adam`

see more details in guide.md [Optimazer section](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/5_Optimazer.md)

- `Layer Norm`

see more details in guide.md [Module section](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/4_Module.md)

- `Batch Norm`

see more details in guide.md [Module section](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/4_Module.md)

- `Dataset` & `DataHandler`

see more details in guide.md [Dataset and DataHandler section](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/7_Dataset_and_DataHandler.md)


### 🚀 Optimizations
- implemented SIMD for matmul

### 🛠️ Bug Fixes
- Broadcast error during scalar operation
- Broadcast error when [x] will be broadcast to [1, x]


# 📦 Version: `0.0.4`

### ✨ New Features
- `Dropout`
- `SGD + Momentum`
- `AdaGrad`
- `powf`
- `train` and `val` method for [module](https://github.com/araxnoid-code/ROTTA-rs/blob/main/book/section/4_Module.md)
- New method for creating `tensors`


### 🚀 Optimizations
- Optimized Basic Operations `add`, `sub`, `mul`, `div`, `matmul`

### 🛠️ Bug Fixes
- Fixed bug on `Sum Square Residual`
- Fixed a bug where tensors accumulated their gradients

# version 0.0.3
### New
- negative indexing for sum_axis, slice, indexing, reshape.
- transpose
- reshape
- to_shape
- slice
- permute
- requires_grad
- Mean Absolute Error
- Mean Square Error
- sign
- abs
- sum function

### Optimalization
- optimizing operations with scalars

### fix bug
- fixed a bug in basic math in arrayy
- fixed backward error in multiple tensor

# version 0.0.2
### New
- Softplus
- ln
- powi
- sigmoid
- mul operation for tensor
- sub operation for tensor
- rename 'reshape' method to 'to_shape' in Arrayy
- update the algorithm of cross entropy loss
- update the algorithm of indexing on Arrayy

### fix bug
- fix bug in tensor broadcasting
- fix bug in derivative of divided for tensor

# version 0.0.1
tensor
- powered by arrayy(look in the folder with the name arrayy)

optimazer
- SGD

loss function
- Sum Square Residual
- Cross Entropy Loss

activation function
- Relu
- Softmax

module
- linear function
- has 3 weight initialization methods(Random, Glorot(default), He)

How to change weight initialization:
```rust
mod rotta_rs;

fn main() {
    let mut model = Module::init();
    model.update_initialization(WeightInitialization::He);
}
```
- default seed is 42

How to change module seed:
```rust
mod rotta_rs;

fn main() {
    let mut model = Module::init();
    model.update_seed(43);
}
```