Training in progress, epoch 0

config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "ingeol/kosaul_v0.2",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 8192,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.40.1",
+  "use_cache": false,
+  "vocab_size": 128256
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbad7a51936e9d60cc0f7618e20fe0a56e387481ec70f21cb5b60f23b6a7a931
+size 6651

upload.py

@@ -0,0 +1,16 @@
+from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+ckpt = 'checkpoint-70'
+origin_model_id = 'beomi/Llama-3-Open-Ko-8B'
+model = AutoModelForCausalLM.from_pretrained(ckpt)
+tokenizer = AutoTokenizer.from_pretrained(origin_model_id)
+
+tokenizer.pad_token = tokenizer.eos_token
+model.resize_token_embeddings(len(tokenizer))
+
+# state_dict = get_fp32_state_dict_from_zero_checkpoint(ckpt)
+# model.load_state_dict(state_dict)
+
+tokenizer.push_to_hub('ingeol/kosaul_v0.2')
+model.push_to_hub('ingeol/kosaul_v0.2')