Worker To Worker Communication (#342)

* [W2WComm] Updates * [W2W_Com] Added handle_w2w_msg function in clientStatem at all states * [W2W] Fed Exp Jsons * [W2W] WIP * [W2W] WIP * [W2W] Updated Jsons * [W2W] Independent Exp Works * [W2W] Added DC Dist_Fed Json * [W2W] Testing Distributed * [W2W] Added start/end_stream messages * [W2W] WIP * [W2W] WIP * [W2W] Distributed Exp WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] Fixed bugs * [W2W] Fixed bugs * [W2W] Fixed bugs * [W2W] Fixed bugs * [W2W] Fixed bugs * [W2W] Fixed bugs * [W2W] Fixed bugs * [W2W] Predict phase update * [W2W] Fix bug * [W2W] Done * [W2W] Fixed averaging * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] Added worker_done * [W2W] Fixed var * [W2W] Fixed var * [W2W] Test * [W2W] Test * [W2W] Test * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] WIP * [W2W] Test * [W2W] Fixed end_stream while state wait * [W2W] Done * [W2W] Done --------- Co-authored-by: David <[email protected]>
leondavi · May 22, 2024 · 72104e7 · 72104e7
1 parent 1a3612d
commit 72104e7
Show file tree

Hide file tree

Showing 28 changed files with 1,529 additions and 396 deletions.
diff --git a/inputJsonsFiles/ConnectionMap/conn_fed_dist_2d_3c_2s_3r_6w.json b/inputJsonsFiles/ConnectionMap/conn_fed_dist_2d_3c_2s_3r_6w.json
@@ -0,0 +1,8 @@
+{
+    "connectionsMap":
+    {
+        "r1":["mainServer", "r2" , "c2" , "r3"],
+        "r2":["r1", "s1" , "c1" , "r3"],
+        "r3":["r1", "r2" , "s2" , "c3"]
+    }
+}
diff --git a/inputJsonsFiles/ConnectionMap/conn_fed_synt_1d_2c_2r_1s_4w_1ws.json b/inputJsonsFiles/ConnectionMap/conn_fed_synt_1d_2c_2r_1s_4w_1ws.json
@@ -0,0 +1,7 @@
+{
+    "connectionsMap":
+    {
+        "r1":["mainServer", "r2" , "c2"],
+        "r2":["r1", "s1" , "c1"]
+    }
+}
diff --git a/inputJsonsFiles/DistributedConfig/dc_AEC_1d_2c_1s_4r_4w.json b/inputJsonsFiles/DistributedConfig/dc_AEC_1d_2c_1s_4r_4w.json
@@ -0,0 +1,67 @@
+{
+    "nerlnetSettings": {
+        "frequency": "200",
+        "batchSize": "100"
+    },
+    "mainServer": {
+        "port": "8081",
+        "args": ""
+    },
+    "apiServer": {
+        "port": "8082",
+        "args": ""
+    },
+    "devices": [
+        {
+            "name": "pc1",
+            "ipv4": "10.211.55.3",
+            "entities": "c1,c2,r2,r1,r3,r4,s1,apiServer,mainServer"
+        }
+    ],
+    "routers": [
+        {
+            "name": "r1",
+            "port": "8086",
+            "policy": "0"
+        },
+        {
+            "name": "r2",
+            "port": "8087",
+            "policy": "0"
+        },
+        {
+            "name": "r3",
+            "port": "8088",
+            "policy": "0"
+        },
+        {
+            "name": "r4",
+            "port": "8089",
+            "policy": "0"
+        }
+    ],
+    "sources": [
+        {
+            "name": "s1",
+            "port": "8085",
+            "frequency": "200",
+            "policy": "0",
+            "epochs": "1",
+            "type": "0"
+        }
+    ],
+    "clients": [
+        {
+            "name": "c1",
+            "port": "8083",
+            "workers": ""
+        },
+        {
+            "name": "c2",
+            "port": "8084",
+            "workers": ""
+        }
+    ],
+    "workers": [],
+    "model_sha": {}
+}
diff --git a/inputJsonsFiles/DistributedConfig/dc_dist_2d_3c_2s_3r_6w.json b/inputJsonsFiles/DistributedConfig/dc_dist_2d_3c_2s_3r_6w.json
@@ -0,0 +1,143 @@
+{
+    "nerlnetSettings": {
+        "frequency": "100",
+        "batchSize": "100"
+    },
+    "mainServer": {
+        "port": "8900",
+        "args": ""
+    },
+    "apiServer": {
+        "port": "8901",
+        "args": ""
+    },
+    "devices": [
+        {
+            "name": "c0vm0",
+            "ipv4": "10.0.0.5",
+            "entities": "mainServer,c1,c2,r1,r2,s1,apiServer"
+        },
+        {
+            "name": "c0vm1",
+            "ipv4": "10.0.0.4",
+            "entities": "c3,r3,s2"
+        }
+    ],
+    "routers": [
+        {
+            "name": "r1",
+            "port": "8905",
+            "policy": "0"
+        },
+        {
+            "name": "r2",
+            "port": "8906",
+            "policy": "0"
+        },
+        {
+            "name": "r3",
+            "port": "8901",
+            "policy": "0"
+        }
+    ],
+    "sources": [
+        {
+            "name": "s1",
+            "port": "8904",
+            "frequency": "200",
+            "policy": "0",
+            "epochs": "1",
+            "type": "0"
+        },
+        {
+            "name": "s2",
+            "port": "8902",
+            "frequency": "200",
+            "policy": "0",
+            "epochs": "1",
+            "type": "0"
+        }
+    ],
+    "clients": [
+        {
+            "name": "c1",
+            "port": "8902",
+            "workers": "w1,w2,ws"
+        },
+        {
+            "name": "c2",
+            "port": "8903",
+            "workers": "w3,w4"
+        },
+        {
+            "name": "c3",
+            "port": "8900",
+            "workers": "w5,w6"
+        }
+    ],
+    "workers": [
+        {
+            "name": "w1",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        },
+        {
+            "name": "w2",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        },
+        {
+            "name": "ws",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        },
+        {
+            "name": "w3",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        },
+        {
+            "name": "w4",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        },
+        {
+            "name": "w5",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        },
+        {
+            "name": "w6",
+            "model_sha": "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896"
+        }
+    ],
+    "model_sha": {
+        "0771693392e898393c9b2b8235497537b5fbed1fd0c9a5a7ec6aab665d2c1896": {
+            "modelType": "0",
+            "_doc_modelType": " nn:0 | approximation:1 | classification:2 | forecasting:3 | image_classification:4 | text_classification:5 | text_generation:6 | auto_association:7 | autoencoder:8 | ae_classifier:9 |",
+            "modelArgs": "",
+            "_doc_modelArgs": "Extra arguments to model",
+            "layersSizes": "5,6,6,4,3",
+            "_doc_layersSizes": "List of postive integers [L0, L1, ..., LN]",
+            "layerTypesList": "1,3,3,3,3",
+            "_doc_LayerTypes": " Default:0 | Scaling:1 | Conv:2 | Perceptron:3 | Pooling:4 | Probabilistic:5 | LSTM:6 | Reccurrent:7 | Unscaling:8 | Flatten:9 | Bounding:10 |",
+            "layers_functions": "1,8,8,8,11",
+            "_doc_layers_functions_activation": " Threshold:1 | Sign:2 | Logistic:3 | Tanh:4 | Linear:5 | ReLU:6 | eLU:7 | SeLU:8 | Soft-plus:9 | Soft-sign:10 | Hard-sigmoid:11 |",
+            "_doc_layer_functions_pooling": " none:1 | Max:2 | Avg:3 |",
+            "_doc_layer_functions_probabilistic": " Binary:1 | Logistic:2 | Competitive:3 | Softmax:4 |",
+            "_doc_layer_functions_scaler": " none:1 | MinMax:2 | MeanStd:3 | STD:4 | Log:5 |",
+            "lossMethod": "2",
+            "_doc_lossMethod": " SSE:1 | MSE:2 | NSE:3 | MinkowskiE:4 | WSE:5 | CEE:6 |",
+            "lr": "0.001",
+            "_doc_lr": "Positve float",
+            "epochs": "1",
+            "_doc_epochs": "Positve Integer",
+            "optimizer": "5",
+            "_doc_optimizer": " GD:0 | CGD:1 | SGD:2 | QuasiNeuton:3 | LVM:4 | ADAM:5 |",
+            "optimizerArgs": "none",
+            "_doc_optimizerArgs": "String",
+            "infraType": "0",
+            "_doc_infraType": " opennn:0 | wolfengine:1 |",
+            "distributedSystemType": "0",
+            "_doc_distributedSystemType": " none:0 | fedClientAvg:1 | fedServerAvg:2 |",
+            "distributedSystemArgs": "SyncMaxCount=10",
+            "_doc_distributedSystemArgs": "String",
+            "distributedSystemToken": "9922u",
+            "_doc_distributedSystemToken": "Token that associates distributed group of workers and parameter-server"
+        }
+    }
+}