misc

mjc_notes.md

@@ -46,3 +46,20 @@ python -m experiments.forecast --config_path=storage/experiments/Stocks/96S/repe
 make build-all path=experiments/configs/Stocks
 ./run.sh
 ```
+
+# So how does deeptime work?
+
+original:
+- inr(coords)
+- RR
+
+My mods (I added past other variables):
+- inr(concat([x, coords]))
+- RR
+
+Where INR is one of [mlp, lstm, lstm2, transformer, transforme2, inceptioncausal]
+
+TODO:
+- M2S mode
+- add other INR's
+- add None as learner

models/DeepTIMe.py

@@ -36,18 +36,23 @@ class DeepTIMe(nn.Module):
     def forward(self, x: Tensor, x_time: Tensor, y_time: Tensor) -> Tensor:
         tgt_horizon_len = y_time.shape[1]
         batch_size, lookback_len, _ = x.shape
+
+        # relative coordinates are the same for each batch, so we make them once and repeat them        
         coords = self.get_coords(lookback_len, tgt_horizon_len).to(x.device)
 
-        if y_time.shape[-1] != 0:
+        if y_time.shape[-1] != 0: # if y_time is not empty of features
             time = torch.cat([x_time, y_time], dim=1)
             coords = repeat(coords, '1 t 1 -> b t 1', b=time.shape[0])
             coords = torch.cat([coords, time], dim=-1)
             time_reprs = self.inr(coords)
         else:
-            time_reprs = repeat(self.inr(coords), '1 t d -> b t d', b=batch_size)
+            a = self.inr(coords)
+            time_reprs = repeat(a, '1 t d -> b t d', b=batch_size)
+            print(coords.shape, a.shape, batch_size, time_reprs.shape)
 
         lookback_reprs = time_reprs[:, :-tgt_horizon_len]
         horizon_reprs = time_reprs[:, -tgt_horizon_len:]
+        
         w, b = self.adaptive_weights(lookback_reprs, x)
         preds = self.forecast(horizon_reprs, w, b)
         return preds

scratch-single.ipynb

@@ -6,8 +6,8 @@
    "id": "4e09086b",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2022-11-20T05:05:01.730258Z",
-     "start_time": "2022-11-20T05:05:00.558925Z"
+     "end_time": "2022-11-20T10:31:28.624566Z",
+     "start_time": "2022-11-20T10:31:27.336916Z"
     },
     "lines_to_next_cell": 0
    },
@@ -71,8 +71,8 @@
    "id": "04499bef",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2022-11-20T05:05:01.737023Z",
-     "start_time": "2022-11-20T05:05:01.731574Z"
+     "end_time": "2022-11-20T10:31:28.634463Z",
+     "start_time": "2022-11-20T10:31:28.626256Z"
     }
    },
    "outputs": [],
@@ -125,12 +125,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 34,
+   "execution_count": 3,
    "id": "768530be",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2022-11-20T06:52:29.098202Z",
-     "start_time": "2022-11-20T06:52:29.093424Z"
+     "end_time": "2022-11-20T10:31:28.667218Z",
+     "start_time": "2022-11-20T10:31:28.635969Z"
     }
    },
    "outputs": [],
@@ -184,7 +184,7 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "739ee5e3",
+   "id": "f6da82f9",
    "metadata": {
     "ExecuteTime": {
      "end_time": "2022-11-20T06:50:03.264769Z",
@@ -196,12 +196,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 35,
+   "execution_count": 4,
    "id": "b8843217",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2022-11-20T06:52:29.508811Z",
-     "start_time": "2022-11-20T06:52:29.353139Z"
+     "end_time": "2022-11-20T10:31:32.788856Z",
+     "start_time": "2022-11-20T10:31:28.668594Z"
     }
    },
    "outputs": [
@@ -221,7 +221,7 @@
        "1"
       ]
      },
-     "execution_count": 35,
+     "execution_count": 4,
      "metadata": {},
      "output_type": "execute_result"
     },
@@ -249,12 +249,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 36,
+   "execution_count": 5,
    "id": "d39ba7c5",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2022-11-20T06:52:41.798916Z",
-     "start_time": "2022-11-20T06:52:41.260013Z"
+     "end_time": "2022-11-20T10:31:33.408193Z",
+     "start_time": "2022-11-20T10:31:32.789874Z"
     }
    },
    "outputs": [
@@ -275,7 +275,7 @@
        "1"
       ]
      },
-     "execution_count": 36,
+     "execution_count": 5,
      "metadata": {},
      "output_type": "execute_result"
     },
@@ -301,12 +301,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 14,
+   "execution_count": 6,
    "id": "9af56ddc",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2022-11-20T06:28:51.826377Z",
-     "start_time": "2022-11-20T06:28:51.721432Z"
+     "end_time": "2022-11-20T10:31:33.539388Z",
+     "start_time": "2022-11-20T10:31:33.409217Z"
     }
    },
    "outputs": [