Training in progress, step 20

config.json

@@ -0,0 +1,46 @@
+{
+  "_name_or_path": "ai4bharat/indictrans2-en-indic-dist-200M",
+  "activation_dropout": 0.0,
+  "activation_function": "gelu",
+  "architectures": [
+    "IndicTransForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "attn_implementation": null,
+  "auto_map": {
+    "AutoConfig": "ai4bharat/indictrans2-en-indic-dist-200M--configuration_indictrans.IndicTransConfig",
+    "AutoModelForSeq2SeqLM": "ai4bharat/indictrans2-en-indic-dist-200M--modeling_indictrans.IndicTransForConditionalGeneration"
+  },
+  "bos_token_id": 0,
+  "decoder_attention_heads": 8,
+  "decoder_embed_dim": 512,
+  "decoder_ffn_dim": 2048,
+  "decoder_layerdrop": 0,
+  "decoder_layers": 18,
+  "decoder_normalize_before": true,
+  "decoder_start_token_id": 2,
+  "decoder_vocab_size": 122672,
+  "dropout": 0.2,
+  "encoder_attention_heads": 8,
+  "encoder_embed_dim": 512,
+  "encoder_ffn_dim": 2048,
+  "encoder_layerdrop": 0,
+  "encoder_layers": 18,
+  "encoder_normalize_before": true,
+  "encoder_vocab_size": 32322,
+  "eos_token_id": 2,
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "layernorm_embedding": true,
+  "max_source_positions": 256,
+  "max_target_positions": 256,
+  "model_type": "IndicTrans",
+  "num_hidden_layers": 18,
+  "pad_token_id": 1,
+  "scale_embedding": true,
+  "share_decoder_input_output_embed": true,
+  "tokenizer_class": "IndicTransTokenizer",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.44.0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f737aa413837de162bcd5c24973bba6f116a1736dcc653dea3ff74eb0a5b887e
+size 549258952

runs/Aug23_00-32-05_machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC/events.out.tfevents.1724353325.machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC.112323.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86b95db0fd32b4ae8e41eaeebe8fb54d5537346a81ad4c971d3e96960d0cfb07
+size 48103

runs/Aug23_15-04-56_machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC/events.out.tfevents.1724405697.machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC.116698.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9614ba8a9045b291595d451131fc7dd689f3c844ea39a25f80987efcdf49c16f
+size 5919

runs/Aug23_15-06-53_machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC/events.out.tfevents.1724405814.machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC.116769.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42a584fc57f15a7ab6b88645a16fe783a36312772f287ac1b174a29b1a7df2fa
+size 5920

runs/Aug23_15-08-23_machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC/events.out.tfevents.1724405904.machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC.116830.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6fe5a11cedc3b54ea0d4041b5ecd6e088180280f91095399268ec28d8627215
+size 5921

runs/Aug23_15-09-37_machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC/events.out.tfevents.1724405977.machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC.116889.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67bdfe8583aebed7aabd45f55e9453e02b99fb5261e9467d0d50b36e7a2c647b
+size 5921

runs/Aug23_15-11-41_machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC/events.out.tfevents.1724406102.machine1-HP-Z2-Tower-G9-Workstation-Desktop-PC.116967.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d6a3fcaf75cb091c242e56e397f98eebd86c3769a45abc917bf79b2f0898220
+size 6452

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1b9de20ea5e9bcbac44cfdc5a7eceab986294acf3257afd6b2a9ec119750b33
+size 5368