diff --git a/checkpoint-150/config.json b/checkpoint-150/config.json
deleted file mode 100644
index 82bacd02f94ab4a4cfdb7b12a0484fac8f301916..0000000000000000000000000000000000000000
--- a/checkpoint-150/config.json
+++ /dev/null
@@ -1,25 +0,0 @@
-{
-  "_name_or_path": "mistralai/Mistral-7B-v0.1",
-  "architectures": [
-    "MistralForCausalLM"
-  ],
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "hidden_act": "silu",
-  "hidden_size": 4096,
-  "initializer_range": 0.02,
-  "intermediate_size": 14336,
-  "max_position_embeddings": 32768,
-  "model_type": "mistral",
-  "num_attention_heads": 32,
-  "num_hidden_layers": 32,
-  "num_key_value_heads": 8,
-  "rms_norm_eps": 1e-05,
-  "rope_theta": 10000.0,
-  "sliding_window": 4096,
-  "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
-  "transformers_version": "4.34.0.dev0",
-  "use_cache": false,
-  "vocab_size": 32002
-}
diff --git a/checkpoint-150/generation_config.json b/checkpoint-150/generation_config.json
deleted file mode 100644
index 2c5f418036a121b3fd432d1bf2b3c5c9daf59fab..0000000000000000000000000000000000000000
--- a/checkpoint-150/generation_config.json
+++ /dev/null
@@ -1,6 +0,0 @@
-{
-  "_from_model_config": true,
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "transformers_version": "4.34.0.dev0"
-}
diff --git a/checkpoint-150/latest b/checkpoint-150/latest
deleted file mode 100644
index daf5be2c4861b36c6659b05fae8c31547db7f579..0000000000000000000000000000000000000000
--- a/checkpoint-150/latest
+++ /dev/null
@@ -1 +0,0 @@
-global_step150
\ No newline at end of file
diff --git a/checkpoint-150/pytorch_model-00001-of-00002.bin b/checkpoint-150/pytorch_model-00001-of-00002.bin
deleted file mode 100644
index ad9d9c0216e08e06631df3ed167f709cb5f91f70..0000000000000000000000000000000000000000
--- a/checkpoint-150/pytorch_model-00001-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5dfffab3ff404e1f12ab78eeaa64de020182e23ef1fdf655b67f138f53b57776
-size 9943044428
diff --git a/checkpoint-150/pytorch_model-00002-of-00002.bin b/checkpoint-150/pytorch_model-00002-of-00002.bin
deleted file mode 100644
index dde4607f61bde844207c2e106c839555d2c48d06..0000000000000000000000000000000000000000
--- a/checkpoint-150/pytorch_model-00002-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b5d5ce3c93e1b594fb0c685762baf81fb339d78111f6fe4459084b45d5dbc36d
-size 4540552031
diff --git a/checkpoint-150/pytorch_model.bin.index.json b/checkpoint-150/pytorch_model.bin.index.json
deleted file mode 100644
index 53213fb82ddc02718be2ce686f00ba7fb0af95e7..0000000000000000000000000000000000000000
--- a/checkpoint-150/pytorch_model.bin.index.json
+++ /dev/null
@@ -1,298 +0,0 @@
-{
-  "metadata": {
-    "total_size": 14483496960
-  },
-  "weight_map": {
-    "lm_head.weight": "pytorch_model-00002-of-00002.bin",
-    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.23.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.norm.weight": "pytorch_model-00002-of-00002.bin"
-  }
-}
diff --git a/checkpoint-150/rng_state_0.pth b/checkpoint-150/rng_state_0.pth
deleted file mode 100644
index 24aba7bcc2a9fb783bd13a6a67b96e5ad055d89d..0000000000000000000000000000000000000000
--- a/checkpoint-150/rng_state_0.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1eafe3d5e0585dde8c5033613de99a5d4f23df4284a488f4007b3944580c0b97
-size 17655
diff --git a/checkpoint-150/rng_state_1.pth b/checkpoint-150/rng_state_1.pth
deleted file mode 100644
index 6b2ef88173fde17f2b3e738a28446f89a0528a96..0000000000000000000000000000000000000000
--- a/checkpoint-150/rng_state_1.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7e34eb456d2d003a2839f2daa9425e99bdd79ed7e24a1de9fc7d5738476bfb4b
-size 17655
diff --git a/checkpoint-150/rng_state_2.pth b/checkpoint-150/rng_state_2.pth
deleted file mode 100644
index 7b118d52a3006aea6c44f23f94c5568d1fb0a2f3..0000000000000000000000000000000000000000
--- a/checkpoint-150/rng_state_2.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b374af4a2765d8771cee7a72921d3c2e438b9bee34f0b2d098ce6071afeb65e4
-size 17655
diff --git a/checkpoint-150/rng_state_3.pth b/checkpoint-150/rng_state_3.pth
deleted file mode 100644
index 3f6fd9aa58eb1d5815ca991134531a3280601900..0000000000000000000000000000000000000000
--- a/checkpoint-150/rng_state_3.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5df75d8477fcc69c7abb03025313915ebfe3ac18c54a7c57aaa455c0099e13e5
-size 17655
diff --git a/checkpoint-150/trainer_state.json b/checkpoint-150/trainer_state.json
deleted file mode 100644
index be94d2c7da8a5a56e771ef2733e94c9c2fc2730a..0000000000000000000000000000000000000000
--- a/checkpoint-150/trainer_state.json
+++ /dev/null
@@ -1,943 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.019846520243450648,
-  "eval_steps": 756,
-  "global_step": 150,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 0.9197,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_loss": 1.4652303457260132,
-      "eval_runtime": 2.1726,
-      "eval_samples_per_second": 79.627,
-      "eval_steps_per_second": 3.682,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_bench_accuracy_agieval": 0.2711864406779661,
-      "eval_bench_accuracy_arc_challenge": 0.8703703703703703,
-      "eval_bench_accuracy_arc_easy": 0.9259259259259259,
-      "eval_bench_accuracy_bigbench": 0.36065573770491804,
-      "eval_bench_accuracy_boolq": 0.5740740740740741,
-      "eval_bench_accuracy_mmlu": 0.5185185185185185,
-      "eval_bench_accuracy_openbookqa": 0.1111111111111111,
-      "eval_bench_accuracy_truthful_qa": 0.3584905660377358,
-      "eval_bench_accuracy_winogrande": 0.4444444444444444,
-      "eval_bench_average_accuracy": 0.4927530209850072,
-      "eval_bench_loss": 2.6978388407144203,
-      "eval_bench_total_accuracy": 0.48893360160965793,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 1.3426,
-      "step": 2
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 1.5882,
-      "step": 3
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.8e-06,
-      "loss": 0.8542,
-      "step": 4
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 0.9629,
-      "step": 5
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3e-06,
-      "loss": 0.903,
-      "step": 6
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.6e-06,
-      "loss": 0.909,
-      "step": 7
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.2e-06,
-      "loss": 0.8666,
-      "step": 8
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 1.0108,
-      "step": 9
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.4e-06,
-      "loss": 0.8958,
-      "step": 10
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6e-06,
-      "loss": 0.9348,
-      "step": 11
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999602806831722e-06,
-      "loss": 0.7832,
-      "step": 12
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999205613663445e-06,
-      "loss": 0.8083,
-      "step": 13
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9988084204951675e-06,
-      "loss": 0.8164,
-      "step": 14
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99841122732689e-06,
-      "loss": 0.7834,
-      "step": 15
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.998014034158613e-06,
-      "loss": 0.8718,
-      "step": 16
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997616840990336e-06,
-      "loss": 0.84,
-      "step": 17
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997219647822058e-06,
-      "loss": 0.7397,
-      "step": 18
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99682245465378e-06,
-      "loss": 0.7445,
-      "step": 19
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996425261485502e-06,
-      "loss": 0.7898,
-      "step": 20
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996028068317225e-06,
-      "loss": 0.7388,
-      "step": 21
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9956308751489475e-06,
-      "loss": 0.7296,
-      "step": 22
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99523368198067e-06,
-      "loss": 0.7993,
-      "step": 23
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994836488812393e-06,
-      "loss": 0.7188,
-      "step": 24
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994439295644115e-06,
-      "loss": 0.7473,
-      "step": 25
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994042102475838e-06,
-      "loss": 0.6997,
-      "step": 26
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99364490930756e-06,
-      "loss": 0.725,
-      "step": 27
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.993247716139283e-06,
-      "loss": 0.7272,
-      "step": 28
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992850522971005e-06,
-      "loss": 0.7427,
-      "step": 29
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992453329802727e-06,
-      "loss": 0.7309,
-      "step": 30
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99205613663445e-06,
-      "loss": 0.6764,
-      "step": 31
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991658943466173e-06,
-      "loss": 0.7556,
-      "step": 32
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991261750297895e-06,
-      "loss": 0.7301,
-      "step": 33
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990864557129617e-06,
-      "loss": 0.6776,
-      "step": 34
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99046736396134e-06,
-      "loss": 0.6884,
-      "step": 35
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990070170793063e-06,
-      "loss": 0.7179,
-      "step": 36
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.989672977624785e-06,
-      "loss": 0.6915,
-      "step": 37
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.989275784456507e-06,
-      "loss": 0.7308,
-      "step": 38
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98887859128823e-06,
-      "loss": 0.6743,
-      "step": 39
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9884813981199526e-06,
-      "loss": 0.6604,
-      "step": 40
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.988084204951675e-06,
-      "loss": 0.6609,
-      "step": 41
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.987687011783397e-06,
-      "loss": 0.6524,
-      "step": 42
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98728981861512e-06,
-      "loss": 0.6386,
-      "step": 43
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986892625446843e-06,
-      "loss": 0.728,
-      "step": 44
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986495432278565e-06,
-      "loss": 0.6971,
-      "step": 45
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986098239110287e-06,
-      "loss": 0.6772,
-      "step": 46
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98570104594201e-06,
-      "loss": 0.6774,
-      "step": 47
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9853038527737325e-06,
-      "loss": 0.6868,
-      "step": 48
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984906659605455e-06,
-      "loss": 0.7169,
-      "step": 49
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984509466437178e-06,
-      "loss": 0.669,
-      "step": 50
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9841122732689e-06,
-      "loss": 0.7112,
-      "step": 51
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.983715080100622e-06,
-      "loss": 0.6667,
-      "step": 52
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.983317886932344e-06,
-      "loss": 0.6528,
-      "step": 53
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.982920693764068e-06,
-      "loss": 0.6699,
-      "step": 54
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98252350059579e-06,
-      "loss": 0.6584,
-      "step": 55
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9821263074275125e-06,
-      "loss": 0.6328,
-      "step": 56
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.981729114259235e-06,
-      "loss": 0.6472,
-      "step": 57
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.981331921090958e-06,
-      "loss": 0.6992,
-      "step": 58
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98093472792268e-06,
-      "loss": 0.6666,
-      "step": 59
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.980537534754402e-06,
-      "loss": 0.6819,
-      "step": 60
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.980140341586125e-06,
-      "loss": 0.705,
-      "step": 61
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.979743148417847e-06,
-      "loss": 0.6871,
-      "step": 62
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97934595524957e-06,
-      "loss": 0.6998,
-      "step": 63
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.978948762081292e-06,
-      "loss": 0.6081,
-      "step": 64
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9785515689130154e-06,
-      "loss": 0.6985,
-      "step": 65
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.978154375744738e-06,
-      "loss": 0.6631,
-      "step": 66
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97775718257646e-06,
-      "loss": 0.6534,
-      "step": 67
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.977359989408182e-06,
-      "loss": 0.6685,
-      "step": 68
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976962796239905e-06,
-      "loss": 0.6821,
-      "step": 69
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976565603071627e-06,
-      "loss": 0.6241,
-      "step": 70
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976168409903349e-06,
-      "loss": 0.6357,
-      "step": 71
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.975771216735072e-06,
-      "loss": 0.6466,
-      "step": 72
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.975374023566795e-06,
-      "loss": 0.6579,
-      "step": 73
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9749768303985176e-06,
-      "loss": 0.6298,
-      "step": 74
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97457963723024e-06,
-      "loss": 0.703,
-      "step": 75
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.974182444061963e-06,
-      "loss": 0.6152,
-      "step": 76
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.973785250893685e-06,
-      "loss": 0.6682,
-      "step": 77
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.973388057725407e-06,
-      "loss": 0.6427,
-      "step": 78
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.972990864557129e-06,
-      "loss": 0.6969,
-      "step": 79
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.972593671388852e-06,
-      "loss": 0.6619,
-      "step": 80
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9721964782205745e-06,
-      "loss": 0.6332,
-      "step": 81
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9717992850522975e-06,
-      "loss": 0.6203,
-      "step": 82
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97140209188402e-06,
-      "loss": 0.6463,
-      "step": 83
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.971004898715743e-06,
-      "loss": 0.6718,
-      "step": 84
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.970607705547465e-06,
-      "loss": 0.6495,
-      "step": 85
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.970210512379187e-06,
-      "loss": 0.5787,
-      "step": 86
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96981331921091e-06,
-      "loss": 0.6897,
-      "step": 87
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.969416126042632e-06,
-      "loss": 0.6688,
-      "step": 88
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9690189328743544e-06,
-      "loss": 0.6697,
-      "step": 89
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.968621739706077e-06,
-      "loss": 0.6156,
-      "step": 90
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9682245465378e-06,
-      "loss": 0.6301,
-      "step": 91
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967827353369523e-06,
-      "loss": 0.6121,
-      "step": 92
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967430160201245e-06,
-      "loss": 0.6177,
-      "step": 93
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967032967032967e-06,
-      "loss": 0.611,
-      "step": 94
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96663577386469e-06,
-      "loss": 0.6359,
-      "step": 95
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.966238580696412e-06,
-      "loss": 0.6417,
-      "step": 96
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.965841387528134e-06,
-      "loss": 0.6312,
-      "step": 97
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.965444194359857e-06,
-      "loss": 0.6184,
-      "step": 98
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9650470011915796e-06,
-      "loss": 0.6724,
-      "step": 99
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.964649808023302e-06,
-      "loss": 0.6833,
-      "step": 100
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.964252614855025e-06,
-      "loss": 0.6433,
-      "step": 101
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.963855421686747e-06,
-      "loss": 0.6766,
-      "step": 102
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96345822851847e-06,
-      "loss": 0.6527,
-      "step": 103
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.963061035350192e-06,
-      "loss": 0.5982,
-      "step": 104
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.962663842181914e-06,
-      "loss": 0.6749,
-      "step": 105
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.962266649013637e-06,
-      "loss": 0.6494,
-      "step": 106
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9618694558453595e-06,
-      "loss": 0.6998,
-      "step": 107
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.961472262677082e-06,
-      "loss": 0.6112,
-      "step": 108
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.961075069508805e-06,
-      "loss": 0.624,
-      "step": 109
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.960677876340528e-06,
-      "loss": 0.6329,
-      "step": 110
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96028068317225e-06,
-      "loss": 0.6491,
-      "step": 111
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.959883490003972e-06,
-      "loss": 0.6672,
-      "step": 112
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.959486296835694e-06,
-      "loss": 0.6279,
-      "step": 113
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.959089103667417e-06,
-      "loss": 0.6479,
-      "step": 114
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9586919104991395e-06,
-      "loss": 0.6214,
-      "step": 115
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.958294717330862e-06,
-      "loss": 0.6618,
-      "step": 116
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957897524162585e-06,
-      "loss": 0.6703,
-      "step": 117
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957500330994307e-06,
-      "loss": 0.6417,
-      "step": 118
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957103137826029e-06,
-      "loss": 0.631,
-      "step": 119
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.956705944657752e-06,
-      "loss": 0.6169,
-      "step": 120
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.956308751489475e-06,
-      "loss": 0.6521,
-      "step": 121
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955911558321197e-06,
-      "loss": 0.6635,
-      "step": 122
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955514365152919e-06,
-      "loss": 0.6496,
-      "step": 123
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955117171984642e-06,
-      "loss": 0.6431,
-      "step": 124
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.954719978816365e-06,
-      "loss": 0.6246,
-      "step": 125
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.954322785648087e-06,
-      "loss": 0.6557,
-      "step": 126
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953925592479809e-06,
-      "loss": 0.6082,
-      "step": 127
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953528399311532e-06,
-      "loss": 0.5941,
-      "step": 128
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953131206143255e-06,
-      "loss": 0.6566,
-      "step": 129
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.952734012974977e-06,
-      "loss": 0.6243,
-      "step": 130
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.952336819806699e-06,
-      "loss": 0.594,
-      "step": 131
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.951939626638422e-06,
-      "loss": 0.68,
-      "step": 132
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9515424334701446e-06,
-      "loss": 0.6302,
-      "step": 133
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.951145240301867e-06,
-      "loss": 0.6251,
-      "step": 134
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.950748047133589e-06,
-      "loss": 0.6326,
-      "step": 135
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.950350853965312e-06,
-      "loss": 0.6314,
-      "step": 136
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949953660797034e-06,
-      "loss": 0.6598,
-      "step": 137
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949556467628757e-06,
-      "loss": 0.6583,
-      "step": 138
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949159274460479e-06,
-      "loss": 0.6162,
-      "step": 139
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.948762081292202e-06,
-      "loss": 0.7042,
-      "step": 140
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9483648881239245e-06,
-      "loss": 0.6733,
-      "step": 141
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.947967694955647e-06,
-      "loss": 0.6103,
-      "step": 142
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94757050178737e-06,
-      "loss": 0.6269,
-      "step": 143
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.947173308619092e-06,
-      "loss": 0.663,
-      "step": 144
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.946776115450814e-06,
-      "loss": 0.5794,
-      "step": 145
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.946378922282537e-06,
-      "loss": 0.6868,
-      "step": 146
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.945981729114259e-06,
-      "loss": 0.6064,
-      "step": 147
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.945584535945982e-06,
-      "loss": 0.6519,
-      "step": 148
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9451873427777045e-06,
-      "loss": 0.655,
-      "step": 149
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.944790149609427e-06,
-      "loss": 0.6617,
-      "step": 150
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 15116,
-  "num_train_epochs": 2,
-  "save_steps": 50,
-  "total_flos": 6.291064218451968e+17,
-  "trial_name": null,
-  "trial_params": null
-}
diff --git a/checkpoint-150/training_args.bin b/checkpoint-150/training_args.bin
deleted file mode 100644
index 731dabe55350d521ab0dfde0b2f023771c347250..0000000000000000000000000000000000000000
--- a/checkpoint-150/training_args.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2f05be88d930176935da1678b48a8294634889bf7ae4f8bebdbaca140c2dac08
-size 5947
diff --git a/checkpoint-150/zero_to_fp32.py b/checkpoint-150/zero_to_fp32.py
deleted file mode 100755
index c98caae31534368be22b67fc4ae906836c992a8d..0000000000000000000000000000000000000000
--- a/checkpoint-150/zero_to_fp32.py
+++ /dev/null
@@ -1,587 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright (c) Microsoft Corporation.
-# SPDX-License-Identifier: Apache-2.0
-
-# DeepSpeed Team
-
-# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
-# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
-# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
-# application.
-#
-# example: python zero_to_fp32.py . pytorch_model.bin
-
-import argparse
-import torch
-import glob
-import math
-import os
-import re
-from collections import OrderedDict
-from dataclasses import dataclass
-
-# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
-# DeepSpeed data structures it has to be available in the current python environment.
-from deepspeed.utils import logger
-from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
-                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
-                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
-
-
-@dataclass
-class zero_model_state:
-    buffers: dict()
-    param_shapes: dict()
-    shared_params: list
-    ds_version: int
-    frozen_param_shapes: dict()
-    frozen_param_fragments: dict()
-
-
-debug = 0
-
-# load to cpu
-device = torch.device('cpu')
-
-
-def atoi(text):
-    return int(text) if text.isdigit() else text
-
-
-def natural_keys(text):
-    '''
-    alist.sort(key=natural_keys) sorts in human order
-    http://nedbatchelder.com/blog/200712/human_sorting.html
-    (See Toothy's implementation in the comments)
-    '''
-    return [atoi(c) for c in re.split(r'(\d+)', text)]
-
-
-def get_model_state_file(checkpoint_dir, zero_stage):
-    if not os.path.isdir(checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
-
-    # there should be only one file
-    if zero_stage <= 2:
-        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
-    elif zero_stage == 3:
-        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
-
-    if not os.path.exists(file):
-        raise FileNotFoundError(f"can't find model states file at '{file}'")
-
-    return file
-
-
-def get_checkpoint_files(checkpoint_dir, glob_pattern):
-    # XXX: need to test that this simple glob rule works for multi-node setup too
-    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
-
-    if len(ckpt_files) == 0:
-        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
-
-    return ckpt_files
-
-
-def get_optim_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
-
-
-def get_model_state_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
-
-
-def parse_model_states(files):
-    zero_model_states = []
-    for file in files:
-        state_dict = torch.load(file, map_location=device)
-
-        if BUFFER_NAMES not in state_dict:
-            raise ValueError(f"{file} is not a model state checkpoint")
-        buffer_names = state_dict[BUFFER_NAMES]
-        if debug:
-            print("Found buffers:", buffer_names)
-
-        # recover just the buffers while restoring them to fp32 if they were saved in fp16
-        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
-        param_shapes = state_dict[PARAM_SHAPES]
-
-        # collect parameters that are included in param_shapes
-        param_names = []
-        for s in param_shapes:
-            for name in s.keys():
-                param_names.append(name)
-
-        # update with frozen parameters
-        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
-        if frozen_param_shapes is not None:
-            if debug:
-                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
-            param_names += list(frozen_param_shapes.keys())
-
-        # handle shared params
-        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
-
-        ds_version = state_dict.get(DS_VERSION, None)
-
-        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
-
-        z_model_state = zero_model_state(buffers=buffers,
-                                         param_shapes=param_shapes,
-                                         shared_params=shared_params,
-                                         ds_version=ds_version,
-                                         frozen_param_shapes=frozen_param_shapes,
-                                         frozen_param_fragments=frozen_param_fragments)
-        zero_model_states.append(z_model_state)
-
-    return zero_model_states
-
-
-def parse_optim_states(files, ds_checkpoint_dir):
-
-    total_files = len(files)
-    state_dicts = []
-    for f in files:
-        state_dict = torch.load(f, map_location=device)
-        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
-        # and also handle the case where it was already removed by another helper script
-        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
-        state_dicts.append(state_dict)
-
-    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
-        raise ValueError(f"{files[0]} is not a zero checkpoint")
-    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
-    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
-
-    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
-    # parameters can be different from data parallelism for non-expert parameters. So we can just
-    # use the max of the partition_count to get the dp world_size.
-
-    if type(world_size) is list:
-        world_size = max(world_size)
-
-    if world_size != total_files:
-        raise ValueError(
-            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
-            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
-        )
-
-    # the groups are named differently in each stage
-    if zero_stage <= 2:
-        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
-    elif zero_stage == 3:
-        fp32_groups_key = FP32_FLAT_GROUPS
-    else:
-        raise ValueError(f"unknown zero stage {zero_stage}")
-
-    if zero_stage <= 2:
-        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
-    elif zero_stage == 3:
-        # if there is more than one param group, there will be multiple flattened tensors - one
-        # flattened tensor per group - for simplicity merge them into a single tensor
-        #
-        # XXX: could make the script more memory efficient for when there are multiple groups - it
-        # will require matching the sub-lists of param_shapes for each param group flattened tensor
-
-        fp32_flat_groups = [
-            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
-        ]
-
-    return zero_stage, world_size, fp32_flat_groups
-
-
-def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
-    """
-    Returns fp32 state_dict reconstructed from ds checkpoint
-
-    Args:
-        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
-
-    """
-    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
-
-    optim_files = get_optim_files(ds_checkpoint_dir)
-    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
-    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
-
-    model_files = get_model_state_files(ds_checkpoint_dir)
-
-    zero_model_states = parse_model_states(model_files)
-    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
-
-    if zero_stage <= 2:
-        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-    elif zero_stage == 3:
-        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-
-
-def _zero2_merge_frozen_params(state_dict, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
-
-    if debug:
-        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
-        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        state_dict[name] = frozen_param_fragments[name]
-
-        if debug:
-            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-
-    # Reconstruction protocol:
-    #
-    # XXX: document this
-
-    if debug:
-        for i in range(world_size):
-            for j in range(len(fp32_flat_groups[0])):
-                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
-
-    # XXX: memory usage doubles here (zero2)
-    num_param_groups = len(fp32_flat_groups[0])
-    merged_single_partition_of_fp32_groups = []
-    for i in range(num_param_groups):
-        merged_partitions = [sd[i] for sd in fp32_flat_groups]
-        full_single_fp32_vector = torch.cat(merged_partitions, 0)
-        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
-    avail_numel = sum(
-        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
-
-    if debug:
-        wanted_params = sum([len(shapes) for shapes in param_shapes])
-        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
-        # not asserting if there is a mismatch due to possible padding
-        print(f"Have {avail_numel} numels to process.")
-        print(f"Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    total_numel = 0
-    total_params = 0
-    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
-        offset = 0
-        avail_numel = full_single_fp32_vector.numel()
-        for name, shape in shapes.items():
-
-            unpartitioned_numel = shape.numel()
-            total_numel += unpartitioned_numel
-            total_params += 1
-
-            if debug:
-                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
-            offset += unpartitioned_numel
-
-        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
-        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
-        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
-        # live optimizer object, so we are checking that the numbers are within the right range
-        align_to = 2 * world_size
-
-        def zero2_align(x):
-            return align_to * math.ceil(x / align_to)
-
-        if debug:
-            print(f"original offset={offset}, avail_numel={avail_numel}")
-
-        offset = zero2_align(offset)
-        avail_numel = zero2_align(avail_numel)
-
-        if debug:
-            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
-
-        # Sanity check
-        if offset != avail_numel:
-            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero2_merge_frozen_params(state_dict, zero_model_states)
-
-    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def zero3_partitioned_param_info(unpartitioned_numel, world_size):
-    remainder = unpartitioned_numel % world_size
-    padding_numel = (world_size - remainder) if remainder else 0
-    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
-    return partitioned_numel, padding_numel
-
-
-def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    if debug:
-        for i in range(world_size):
-            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
-            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in zero_model_states[0].frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
-        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-    avail_numel = fp32_flat_groups[0].numel() * world_size
-    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
-    # param, re-consolidating each param, while dealing with padding if any
-
-    # merge list of dicts, preserving order
-    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
-
-    if debug:
-        for i in range(world_size):
-            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
-
-        wanted_params = len(param_shapes)
-        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
-        # not asserting if there is a mismatch due to possible padding
-        avail_numel = fp32_flat_groups[0].numel() * world_size
-        print(f"Trainable params: Have {avail_numel} numels to process.")
-        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    offset = 0
-    total_numel = 0
-    total_params = 0
-    for name, shape in param_shapes.items():
-
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-        total_params += 1
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-        # XXX: memory usage doubles here
-        state_dict[name] = torch.cat(
-            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
-            0).narrow(0, 0, unpartitioned_numel).view(shape)
-        offset += partitioned_numel
-
-    offset *= world_size
-
-    # Sanity check
-    if offset != avail_numel:
-        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
-
-    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
-    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
-    via a model hub.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
-
-    Returns:
-        - pytorch ``state_dict``
-
-    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
-    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
-    the checkpoint.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
-        # do the training and checkpoint saving
-        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
-        model = model.cpu() # move to cpu
-        model.load_state_dict(state_dict)
-        # submit to model hub or save the model to share with others
-
-    In this example the ``model`` will no longer be usable in the deepspeed context of the same
-    application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
-
-    """
-    if tag is None:
-        latest_path = os.path.join(checkpoint_dir, 'latest')
-        if os.path.isfile(latest_path):
-            with open(latest_path, 'r') as fd:
-                tag = fd.read().strip()
-        else:
-            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
-
-    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
-
-    if not os.path.isdir(ds_checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
-
-    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
-
-
-def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
-    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-    """
-
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-    print(f"Saving fp32 state dict to {output_file}")
-    torch.save(state_dict, output_file)
-
-
-def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
-    """
-    1. Put the provided model to cpu
-    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
-    3. Load it into the provided model
-
-    Args:
-        - ``model``: the model object to update
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-
-    Returns:
-        - ``model`: modified model
-
-    Make sure you have plenty of CPU memory available before you call this function. If you don't
-    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
-    conveniently placed for you in the checkpoint folder.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
-        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
-        # submit to model hub or save the model to share with others
-
-    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
-    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    """
-    logger.info(f"Extracting fp32 weights")
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-
-    logger.info(f"Overwriting model with fp32 weights")
-    model = model.cpu()
-    model.load_state_dict(state_dict, strict=False)
-
-    return model
-
-
-if __name__ == "__main__":
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument("checkpoint_dir",
-                        type=str,
-                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
-    parser.add_argument(
-        "output_file",
-        type=str,
-        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
-    parser.add_argument("-t",
-                        "--tag",
-                        type=str,
-                        default=None,
-                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
-    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
-    args = parser.parse_args()
-
-    debug = args.debug
-
-    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
diff --git a/checkpoint-200/config.json b/checkpoint-200/config.json
deleted file mode 100644
index 82bacd02f94ab4a4cfdb7b12a0484fac8f301916..0000000000000000000000000000000000000000
--- a/checkpoint-200/config.json
+++ /dev/null
@@ -1,25 +0,0 @@
-{
-  "_name_or_path": "mistralai/Mistral-7B-v0.1",
-  "architectures": [
-    "MistralForCausalLM"
-  ],
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "hidden_act": "silu",
-  "hidden_size": 4096,
-  "initializer_range": 0.02,
-  "intermediate_size": 14336,
-  "max_position_embeddings": 32768,
-  "model_type": "mistral",
-  "num_attention_heads": 32,
-  "num_hidden_layers": 32,
-  "num_key_value_heads": 8,
-  "rms_norm_eps": 1e-05,
-  "rope_theta": 10000.0,
-  "sliding_window": 4096,
-  "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
-  "transformers_version": "4.34.0.dev0",
-  "use_cache": false,
-  "vocab_size": 32002
-}
diff --git a/checkpoint-200/generation_config.json b/checkpoint-200/generation_config.json
deleted file mode 100644
index 2c5f418036a121b3fd432d1bf2b3c5c9daf59fab..0000000000000000000000000000000000000000
--- a/checkpoint-200/generation_config.json
+++ /dev/null
@@ -1,6 +0,0 @@
-{
-  "_from_model_config": true,
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "transformers_version": "4.34.0.dev0"
-}
diff --git a/checkpoint-200/latest b/checkpoint-200/latest
deleted file mode 100644
index 753e24e10f3a2489150f458205cf759fd8b6081f..0000000000000000000000000000000000000000
--- a/checkpoint-200/latest
+++ /dev/null
@@ -1 +0,0 @@
-global_step200
\ No newline at end of file
diff --git a/checkpoint-200/pytorch_model-00001-of-00002.bin b/checkpoint-200/pytorch_model-00001-of-00002.bin
deleted file mode 100644
index 53d0ed7838534f1a38ff5ed2e576180ed16a7519..0000000000000000000000000000000000000000
--- a/checkpoint-200/pytorch_model-00001-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:903412ad8d63a3544a84531bed488838561c60f33953ec8821e76bb9806cdf31
-size 9943044428
diff --git a/checkpoint-200/pytorch_model-00002-of-00002.bin b/checkpoint-200/pytorch_model-00002-of-00002.bin
deleted file mode 100644
index 83013b21614d33a616a27ba384d0f6a55c469355..0000000000000000000000000000000000000000
--- a/checkpoint-200/pytorch_model-00002-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2ee97a78f24972026adf4f389f4fe546b265d44003a9d0533a43de09bc36f2fd
-size 4540552031
diff --git a/checkpoint-200/pytorch_model.bin.index.json b/checkpoint-200/pytorch_model.bin.index.json
deleted file mode 100644
index 53213fb82ddc02718be2ce686f00ba7fb0af95e7..0000000000000000000000000000000000000000
--- a/checkpoint-200/pytorch_model.bin.index.json
+++ /dev/null
@@ -1,298 +0,0 @@
-{
-  "metadata": {
-    "total_size": 14483496960
-  },
-  "weight_map": {
-    "lm_head.weight": "pytorch_model-00002-of-00002.bin",
-    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.23.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.norm.weight": "pytorch_model-00002-of-00002.bin"
-  }
-}
diff --git a/checkpoint-200/rng_state_0.pth b/checkpoint-200/rng_state_0.pth
deleted file mode 100644
index 24aba7bcc2a9fb783bd13a6a67b96e5ad055d89d..0000000000000000000000000000000000000000
--- a/checkpoint-200/rng_state_0.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1eafe3d5e0585dde8c5033613de99a5d4f23df4284a488f4007b3944580c0b97
-size 17655
diff --git a/checkpoint-200/rng_state_1.pth b/checkpoint-200/rng_state_1.pth
deleted file mode 100644
index 6b2ef88173fde17f2b3e738a28446f89a0528a96..0000000000000000000000000000000000000000
--- a/checkpoint-200/rng_state_1.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7e34eb456d2d003a2839f2daa9425e99bdd79ed7e24a1de9fc7d5738476bfb4b
-size 17655
diff --git a/checkpoint-200/rng_state_2.pth b/checkpoint-200/rng_state_2.pth
deleted file mode 100644
index 7b118d52a3006aea6c44f23f94c5568d1fb0a2f3..0000000000000000000000000000000000000000
--- a/checkpoint-200/rng_state_2.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b374af4a2765d8771cee7a72921d3c2e438b9bee34f0b2d098ce6071afeb65e4
-size 17655
diff --git a/checkpoint-200/rng_state_3.pth b/checkpoint-200/rng_state_3.pth
deleted file mode 100644
index 3f6fd9aa58eb1d5815ca991134531a3280601900..0000000000000000000000000000000000000000
--- a/checkpoint-200/rng_state_3.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5df75d8477fcc69c7abb03025313915ebfe3ac18c54a7c57aaa455c0099e13e5
-size 17655
diff --git a/checkpoint-200/trainer_state.json b/checkpoint-200/trainer_state.json
deleted file mode 100644
index 5584be8c62df62d622a91aff21471107f042b883..0000000000000000000000000000000000000000
--- a/checkpoint-200/trainer_state.json
+++ /dev/null
@@ -1,1243 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.02646202699126753,
-  "eval_steps": 756,
-  "global_step": 200,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 0.9197,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_loss": 1.4652303457260132,
-      "eval_runtime": 2.1726,
-      "eval_samples_per_second": 79.627,
-      "eval_steps_per_second": 3.682,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_bench_accuracy_agieval": 0.2711864406779661,
-      "eval_bench_accuracy_arc_challenge": 0.8703703703703703,
-      "eval_bench_accuracy_arc_easy": 0.9259259259259259,
-      "eval_bench_accuracy_bigbench": 0.36065573770491804,
-      "eval_bench_accuracy_boolq": 0.5740740740740741,
-      "eval_bench_accuracy_mmlu": 0.5185185185185185,
-      "eval_bench_accuracy_openbookqa": 0.1111111111111111,
-      "eval_bench_accuracy_truthful_qa": 0.3584905660377358,
-      "eval_bench_accuracy_winogrande": 0.4444444444444444,
-      "eval_bench_average_accuracy": 0.4927530209850072,
-      "eval_bench_loss": 2.6978388407144203,
-      "eval_bench_total_accuracy": 0.48893360160965793,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 1.3426,
-      "step": 2
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 1.5882,
-      "step": 3
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.8e-06,
-      "loss": 0.8542,
-      "step": 4
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 0.9629,
-      "step": 5
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3e-06,
-      "loss": 0.903,
-      "step": 6
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.6e-06,
-      "loss": 0.909,
-      "step": 7
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.2e-06,
-      "loss": 0.8666,
-      "step": 8
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 1.0108,
-      "step": 9
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.4e-06,
-      "loss": 0.8958,
-      "step": 10
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6e-06,
-      "loss": 0.9348,
-      "step": 11
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999602806831722e-06,
-      "loss": 0.7832,
-      "step": 12
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999205613663445e-06,
-      "loss": 0.8083,
-      "step": 13
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9988084204951675e-06,
-      "loss": 0.8164,
-      "step": 14
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99841122732689e-06,
-      "loss": 0.7834,
-      "step": 15
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.998014034158613e-06,
-      "loss": 0.8718,
-      "step": 16
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997616840990336e-06,
-      "loss": 0.84,
-      "step": 17
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997219647822058e-06,
-      "loss": 0.7397,
-      "step": 18
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99682245465378e-06,
-      "loss": 0.7445,
-      "step": 19
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996425261485502e-06,
-      "loss": 0.7898,
-      "step": 20
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996028068317225e-06,
-      "loss": 0.7388,
-      "step": 21
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9956308751489475e-06,
-      "loss": 0.7296,
-      "step": 22
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99523368198067e-06,
-      "loss": 0.7993,
-      "step": 23
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994836488812393e-06,
-      "loss": 0.7188,
-      "step": 24
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994439295644115e-06,
-      "loss": 0.7473,
-      "step": 25
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994042102475838e-06,
-      "loss": 0.6997,
-      "step": 26
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99364490930756e-06,
-      "loss": 0.725,
-      "step": 27
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.993247716139283e-06,
-      "loss": 0.7272,
-      "step": 28
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992850522971005e-06,
-      "loss": 0.7427,
-      "step": 29
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992453329802727e-06,
-      "loss": 0.7309,
-      "step": 30
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99205613663445e-06,
-      "loss": 0.6764,
-      "step": 31
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991658943466173e-06,
-      "loss": 0.7556,
-      "step": 32
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991261750297895e-06,
-      "loss": 0.7301,
-      "step": 33
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990864557129617e-06,
-      "loss": 0.6776,
-      "step": 34
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99046736396134e-06,
-      "loss": 0.6884,
-      "step": 35
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990070170793063e-06,
-      "loss": 0.7179,
-      "step": 36
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.989672977624785e-06,
-      "loss": 0.6915,
-      "step": 37
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.989275784456507e-06,
-      "loss": 0.7308,
-      "step": 38
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98887859128823e-06,
-      "loss": 0.6743,
-      "step": 39
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9884813981199526e-06,
-      "loss": 0.6604,
-      "step": 40
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.988084204951675e-06,
-      "loss": 0.6609,
-      "step": 41
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.987687011783397e-06,
-      "loss": 0.6524,
-      "step": 42
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98728981861512e-06,
-      "loss": 0.6386,
-      "step": 43
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986892625446843e-06,
-      "loss": 0.728,
-      "step": 44
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986495432278565e-06,
-      "loss": 0.6971,
-      "step": 45
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986098239110287e-06,
-      "loss": 0.6772,
-      "step": 46
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98570104594201e-06,
-      "loss": 0.6774,
-      "step": 47
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9853038527737325e-06,
-      "loss": 0.6868,
-      "step": 48
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984906659605455e-06,
-      "loss": 0.7169,
-      "step": 49
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984509466437178e-06,
-      "loss": 0.669,
-      "step": 50
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9841122732689e-06,
-      "loss": 0.7112,
-      "step": 51
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.983715080100622e-06,
-      "loss": 0.6667,
-      "step": 52
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.983317886932344e-06,
-      "loss": 0.6528,
-      "step": 53
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.982920693764068e-06,
-      "loss": 0.6699,
-      "step": 54
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98252350059579e-06,
-      "loss": 0.6584,
-      "step": 55
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9821263074275125e-06,
-      "loss": 0.6328,
-      "step": 56
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.981729114259235e-06,
-      "loss": 0.6472,
-      "step": 57
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.981331921090958e-06,
-      "loss": 0.6992,
-      "step": 58
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98093472792268e-06,
-      "loss": 0.6666,
-      "step": 59
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.980537534754402e-06,
-      "loss": 0.6819,
-      "step": 60
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.980140341586125e-06,
-      "loss": 0.705,
-      "step": 61
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.979743148417847e-06,
-      "loss": 0.6871,
-      "step": 62
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97934595524957e-06,
-      "loss": 0.6998,
-      "step": 63
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.978948762081292e-06,
-      "loss": 0.6081,
-      "step": 64
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9785515689130154e-06,
-      "loss": 0.6985,
-      "step": 65
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.978154375744738e-06,
-      "loss": 0.6631,
-      "step": 66
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97775718257646e-06,
-      "loss": 0.6534,
-      "step": 67
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.977359989408182e-06,
-      "loss": 0.6685,
-      "step": 68
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976962796239905e-06,
-      "loss": 0.6821,
-      "step": 69
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976565603071627e-06,
-      "loss": 0.6241,
-      "step": 70
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976168409903349e-06,
-      "loss": 0.6357,
-      "step": 71
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.975771216735072e-06,
-      "loss": 0.6466,
-      "step": 72
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.975374023566795e-06,
-      "loss": 0.6579,
-      "step": 73
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9749768303985176e-06,
-      "loss": 0.6298,
-      "step": 74
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97457963723024e-06,
-      "loss": 0.703,
-      "step": 75
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.974182444061963e-06,
-      "loss": 0.6152,
-      "step": 76
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.973785250893685e-06,
-      "loss": 0.6682,
-      "step": 77
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.973388057725407e-06,
-      "loss": 0.6427,
-      "step": 78
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.972990864557129e-06,
-      "loss": 0.6969,
-      "step": 79
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.972593671388852e-06,
-      "loss": 0.6619,
-      "step": 80
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9721964782205745e-06,
-      "loss": 0.6332,
-      "step": 81
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9717992850522975e-06,
-      "loss": 0.6203,
-      "step": 82
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97140209188402e-06,
-      "loss": 0.6463,
-      "step": 83
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.971004898715743e-06,
-      "loss": 0.6718,
-      "step": 84
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.970607705547465e-06,
-      "loss": 0.6495,
-      "step": 85
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.970210512379187e-06,
-      "loss": 0.5787,
-      "step": 86
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96981331921091e-06,
-      "loss": 0.6897,
-      "step": 87
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.969416126042632e-06,
-      "loss": 0.6688,
-      "step": 88
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9690189328743544e-06,
-      "loss": 0.6697,
-      "step": 89
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.968621739706077e-06,
-      "loss": 0.6156,
-      "step": 90
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9682245465378e-06,
-      "loss": 0.6301,
-      "step": 91
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967827353369523e-06,
-      "loss": 0.6121,
-      "step": 92
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967430160201245e-06,
-      "loss": 0.6177,
-      "step": 93
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967032967032967e-06,
-      "loss": 0.611,
-      "step": 94
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96663577386469e-06,
-      "loss": 0.6359,
-      "step": 95
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.966238580696412e-06,
-      "loss": 0.6417,
-      "step": 96
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.965841387528134e-06,
-      "loss": 0.6312,
-      "step": 97
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.965444194359857e-06,
-      "loss": 0.6184,
-      "step": 98
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9650470011915796e-06,
-      "loss": 0.6724,
-      "step": 99
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.964649808023302e-06,
-      "loss": 0.6833,
-      "step": 100
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.964252614855025e-06,
-      "loss": 0.6433,
-      "step": 101
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.963855421686747e-06,
-      "loss": 0.6766,
-      "step": 102
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96345822851847e-06,
-      "loss": 0.6527,
-      "step": 103
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.963061035350192e-06,
-      "loss": 0.5982,
-      "step": 104
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.962663842181914e-06,
-      "loss": 0.6749,
-      "step": 105
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.962266649013637e-06,
-      "loss": 0.6494,
-      "step": 106
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9618694558453595e-06,
-      "loss": 0.6998,
-      "step": 107
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.961472262677082e-06,
-      "loss": 0.6112,
-      "step": 108
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.961075069508805e-06,
-      "loss": 0.624,
-      "step": 109
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.960677876340528e-06,
-      "loss": 0.6329,
-      "step": 110
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96028068317225e-06,
-      "loss": 0.6491,
-      "step": 111
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.959883490003972e-06,
-      "loss": 0.6672,
-      "step": 112
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.959486296835694e-06,
-      "loss": 0.6279,
-      "step": 113
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.959089103667417e-06,
-      "loss": 0.6479,
-      "step": 114
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9586919104991395e-06,
-      "loss": 0.6214,
-      "step": 115
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.958294717330862e-06,
-      "loss": 0.6618,
-      "step": 116
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957897524162585e-06,
-      "loss": 0.6703,
-      "step": 117
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957500330994307e-06,
-      "loss": 0.6417,
-      "step": 118
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957103137826029e-06,
-      "loss": 0.631,
-      "step": 119
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.956705944657752e-06,
-      "loss": 0.6169,
-      "step": 120
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.956308751489475e-06,
-      "loss": 0.6521,
-      "step": 121
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955911558321197e-06,
-      "loss": 0.6635,
-      "step": 122
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955514365152919e-06,
-      "loss": 0.6496,
-      "step": 123
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955117171984642e-06,
-      "loss": 0.6431,
-      "step": 124
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.954719978816365e-06,
-      "loss": 0.6246,
-      "step": 125
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.954322785648087e-06,
-      "loss": 0.6557,
-      "step": 126
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953925592479809e-06,
-      "loss": 0.6082,
-      "step": 127
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953528399311532e-06,
-      "loss": 0.5941,
-      "step": 128
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953131206143255e-06,
-      "loss": 0.6566,
-      "step": 129
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.952734012974977e-06,
-      "loss": 0.6243,
-      "step": 130
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.952336819806699e-06,
-      "loss": 0.594,
-      "step": 131
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.951939626638422e-06,
-      "loss": 0.68,
-      "step": 132
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9515424334701446e-06,
-      "loss": 0.6302,
-      "step": 133
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.951145240301867e-06,
-      "loss": 0.6251,
-      "step": 134
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.950748047133589e-06,
-      "loss": 0.6326,
-      "step": 135
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.950350853965312e-06,
-      "loss": 0.6314,
-      "step": 136
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949953660797034e-06,
-      "loss": 0.6598,
-      "step": 137
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949556467628757e-06,
-      "loss": 0.6583,
-      "step": 138
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949159274460479e-06,
-      "loss": 0.6162,
-      "step": 139
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.948762081292202e-06,
-      "loss": 0.7042,
-      "step": 140
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9483648881239245e-06,
-      "loss": 0.6733,
-      "step": 141
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.947967694955647e-06,
-      "loss": 0.6103,
-      "step": 142
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94757050178737e-06,
-      "loss": 0.6269,
-      "step": 143
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.947173308619092e-06,
-      "loss": 0.663,
-      "step": 144
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.946776115450814e-06,
-      "loss": 0.5794,
-      "step": 145
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.946378922282537e-06,
-      "loss": 0.6868,
-      "step": 146
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.945981729114259e-06,
-      "loss": 0.6064,
-      "step": 147
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.945584535945982e-06,
-      "loss": 0.6519,
-      "step": 148
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9451873427777045e-06,
-      "loss": 0.655,
-      "step": 149
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.944790149609427e-06,
-      "loss": 0.6617,
-      "step": 150
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94439295644115e-06,
-      "loss": 0.627,
-      "step": 151
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.943995763272872e-06,
-      "loss": 0.5837,
-      "step": 152
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.943598570104594e-06,
-      "loss": 0.6201,
-      "step": 153
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.943201376936317e-06,
-      "loss": 0.6291,
-      "step": 154
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.942804183768039e-06,
-      "loss": 0.6061,
-      "step": 155
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.942406990599761e-06,
-      "loss": 0.624,
-      "step": 156
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.942009797431484e-06,
-      "loss": 0.6418,
-      "step": 157
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.941612604263207e-06,
-      "loss": 0.5858,
-      "step": 158
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94121541109493e-06,
-      "loss": 0.6407,
-      "step": 159
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.940818217926652e-06,
-      "loss": 0.6222,
-      "step": 160
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.940421024758374e-06,
-      "loss": 0.5938,
-      "step": 161
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.940023831590097e-06,
-      "loss": 0.6157,
-      "step": 162
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.939626638421819e-06,
-      "loss": 0.5989,
-      "step": 163
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.939229445253541e-06,
-      "loss": 0.7056,
-      "step": 164
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.938832252085264e-06,
-      "loss": 0.6606,
-      "step": 165
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9384350589169865e-06,
-      "loss": 0.6303,
-      "step": 166
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9380378657487095e-06,
-      "loss": 0.6332,
-      "step": 167
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.937640672580432e-06,
-      "loss": 0.6197,
-      "step": 168
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.937243479412155e-06,
-      "loss": 0.6318,
-      "step": 169
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.936846286243877e-06,
-      "loss": 0.6598,
-      "step": 170
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.936449093075599e-06,
-      "loss": 0.662,
-      "step": 171
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.936051899907321e-06,
-      "loss": 0.6018,
-      "step": 172
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.935654706739044e-06,
-      "loss": 0.6955,
-      "step": 173
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9352575135707665e-06,
-      "loss": 0.6283,
-      "step": 174
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.934860320402489e-06,
-      "loss": 0.6829,
-      "step": 175
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.934463127234212e-06,
-      "loss": 0.5985,
-      "step": 176
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.934065934065935e-06,
-      "loss": 0.6385,
-      "step": 177
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.933668740897657e-06,
-      "loss": 0.6326,
-      "step": 178
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.933271547729379e-06,
-      "loss": 0.639,
-      "step": 179
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.932874354561102e-06,
-      "loss": 0.6084,
-      "step": 180
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.932477161392824e-06,
-      "loss": 0.6549,
-      "step": 181
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.932079968224546e-06,
-      "loss": 0.6728,
-      "step": 182
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.931682775056269e-06,
-      "loss": 0.6351,
-      "step": 183
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.931285581887992e-06,
-      "loss": 0.6375,
-      "step": 184
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.930888388719714e-06,
-      "loss": 0.6814,
-      "step": 185
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.930491195551437e-06,
-      "loss": 0.5968,
-      "step": 186
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.930094002383159e-06,
-      "loss": 0.6053,
-      "step": 187
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.929696809214882e-06,
-      "loss": 0.6468,
-      "step": 188
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.929299616046604e-06,
-      "loss": 0.6407,
-      "step": 189
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.928902422878326e-06,
-      "loss": 0.6996,
-      "step": 190
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.928505229710049e-06,
-      "loss": 0.6158,
-      "step": 191
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9281080365417716e-06,
-      "loss": 0.6128,
-      "step": 192
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.927710843373494e-06,
-      "loss": 0.6558,
-      "step": 193
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.927313650205216e-06,
-      "loss": 0.6726,
-      "step": 194
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.92691645703694e-06,
-      "loss": 0.6292,
-      "step": 195
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.926519263868662e-06,
-      "loss": 0.6004,
-      "step": 196
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.926122070700384e-06,
-      "loss": 0.599,
-      "step": 197
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.925724877532106e-06,
-      "loss": 0.6374,
-      "step": 198
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.925327684363829e-06,
-      "loss": 0.6472,
-      "step": 199
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9249304911955515e-06,
-      "loss": 0.594,
-      "step": 200
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 15116,
-  "num_train_epochs": 2,
-  "save_steps": 50,
-  "total_flos": 8.388085624602624e+17,
-  "trial_name": null,
-  "trial_params": null
-}
diff --git a/checkpoint-200/training_args.bin b/checkpoint-200/training_args.bin
deleted file mode 100644
index 731dabe55350d521ab0dfde0b2f023771c347250..0000000000000000000000000000000000000000
--- a/checkpoint-200/training_args.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2f05be88d930176935da1678b48a8294634889bf7ae4f8bebdbaca140c2dac08
-size 5947
diff --git a/checkpoint-200/zero_to_fp32.py b/checkpoint-200/zero_to_fp32.py
deleted file mode 100755
index c98caae31534368be22b67fc4ae906836c992a8d..0000000000000000000000000000000000000000
--- a/checkpoint-200/zero_to_fp32.py
+++ /dev/null
@@ -1,587 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright (c) Microsoft Corporation.
-# SPDX-License-Identifier: Apache-2.0
-
-# DeepSpeed Team
-
-# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
-# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
-# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
-# application.
-#
-# example: python zero_to_fp32.py . pytorch_model.bin
-
-import argparse
-import torch
-import glob
-import math
-import os
-import re
-from collections import OrderedDict
-from dataclasses import dataclass
-
-# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
-# DeepSpeed data structures it has to be available in the current python environment.
-from deepspeed.utils import logger
-from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
-                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
-                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
-
-
-@dataclass
-class zero_model_state:
-    buffers: dict()
-    param_shapes: dict()
-    shared_params: list
-    ds_version: int
-    frozen_param_shapes: dict()
-    frozen_param_fragments: dict()
-
-
-debug = 0
-
-# load to cpu
-device = torch.device('cpu')
-
-
-def atoi(text):
-    return int(text) if text.isdigit() else text
-
-
-def natural_keys(text):
-    '''
-    alist.sort(key=natural_keys) sorts in human order
-    http://nedbatchelder.com/blog/200712/human_sorting.html
-    (See Toothy's implementation in the comments)
-    '''
-    return [atoi(c) for c in re.split(r'(\d+)', text)]
-
-
-def get_model_state_file(checkpoint_dir, zero_stage):
-    if not os.path.isdir(checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
-
-    # there should be only one file
-    if zero_stage <= 2:
-        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
-    elif zero_stage == 3:
-        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
-
-    if not os.path.exists(file):
-        raise FileNotFoundError(f"can't find model states file at '{file}'")
-
-    return file
-
-
-def get_checkpoint_files(checkpoint_dir, glob_pattern):
-    # XXX: need to test that this simple glob rule works for multi-node setup too
-    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
-
-    if len(ckpt_files) == 0:
-        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
-
-    return ckpt_files
-
-
-def get_optim_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
-
-
-def get_model_state_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
-
-
-def parse_model_states(files):
-    zero_model_states = []
-    for file in files:
-        state_dict = torch.load(file, map_location=device)
-
-        if BUFFER_NAMES not in state_dict:
-            raise ValueError(f"{file} is not a model state checkpoint")
-        buffer_names = state_dict[BUFFER_NAMES]
-        if debug:
-            print("Found buffers:", buffer_names)
-
-        # recover just the buffers while restoring them to fp32 if they were saved in fp16
-        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
-        param_shapes = state_dict[PARAM_SHAPES]
-
-        # collect parameters that are included in param_shapes
-        param_names = []
-        for s in param_shapes:
-            for name in s.keys():
-                param_names.append(name)
-
-        # update with frozen parameters
-        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
-        if frozen_param_shapes is not None:
-            if debug:
-                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
-            param_names += list(frozen_param_shapes.keys())
-
-        # handle shared params
-        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
-
-        ds_version = state_dict.get(DS_VERSION, None)
-
-        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
-
-        z_model_state = zero_model_state(buffers=buffers,
-                                         param_shapes=param_shapes,
-                                         shared_params=shared_params,
-                                         ds_version=ds_version,
-                                         frozen_param_shapes=frozen_param_shapes,
-                                         frozen_param_fragments=frozen_param_fragments)
-        zero_model_states.append(z_model_state)
-
-    return zero_model_states
-
-
-def parse_optim_states(files, ds_checkpoint_dir):
-
-    total_files = len(files)
-    state_dicts = []
-    for f in files:
-        state_dict = torch.load(f, map_location=device)
-        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
-        # and also handle the case where it was already removed by another helper script
-        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
-        state_dicts.append(state_dict)
-
-    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
-        raise ValueError(f"{files[0]} is not a zero checkpoint")
-    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
-    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
-
-    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
-    # parameters can be different from data parallelism for non-expert parameters. So we can just
-    # use the max of the partition_count to get the dp world_size.
-
-    if type(world_size) is list:
-        world_size = max(world_size)
-
-    if world_size != total_files:
-        raise ValueError(
-            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
-            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
-        )
-
-    # the groups are named differently in each stage
-    if zero_stage <= 2:
-        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
-    elif zero_stage == 3:
-        fp32_groups_key = FP32_FLAT_GROUPS
-    else:
-        raise ValueError(f"unknown zero stage {zero_stage}")
-
-    if zero_stage <= 2:
-        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
-    elif zero_stage == 3:
-        # if there is more than one param group, there will be multiple flattened tensors - one
-        # flattened tensor per group - for simplicity merge them into a single tensor
-        #
-        # XXX: could make the script more memory efficient for when there are multiple groups - it
-        # will require matching the sub-lists of param_shapes for each param group flattened tensor
-
-        fp32_flat_groups = [
-            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
-        ]
-
-    return zero_stage, world_size, fp32_flat_groups
-
-
-def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
-    """
-    Returns fp32 state_dict reconstructed from ds checkpoint
-
-    Args:
-        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
-
-    """
-    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
-
-    optim_files = get_optim_files(ds_checkpoint_dir)
-    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
-    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
-
-    model_files = get_model_state_files(ds_checkpoint_dir)
-
-    zero_model_states = parse_model_states(model_files)
-    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
-
-    if zero_stage <= 2:
-        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-    elif zero_stage == 3:
-        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-
-
-def _zero2_merge_frozen_params(state_dict, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
-
-    if debug:
-        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
-        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        state_dict[name] = frozen_param_fragments[name]
-
-        if debug:
-            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-
-    # Reconstruction protocol:
-    #
-    # XXX: document this
-
-    if debug:
-        for i in range(world_size):
-            for j in range(len(fp32_flat_groups[0])):
-                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
-
-    # XXX: memory usage doubles here (zero2)
-    num_param_groups = len(fp32_flat_groups[0])
-    merged_single_partition_of_fp32_groups = []
-    for i in range(num_param_groups):
-        merged_partitions = [sd[i] for sd in fp32_flat_groups]
-        full_single_fp32_vector = torch.cat(merged_partitions, 0)
-        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
-    avail_numel = sum(
-        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
-
-    if debug:
-        wanted_params = sum([len(shapes) for shapes in param_shapes])
-        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
-        # not asserting if there is a mismatch due to possible padding
-        print(f"Have {avail_numel} numels to process.")
-        print(f"Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    total_numel = 0
-    total_params = 0
-    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
-        offset = 0
-        avail_numel = full_single_fp32_vector.numel()
-        for name, shape in shapes.items():
-
-            unpartitioned_numel = shape.numel()
-            total_numel += unpartitioned_numel
-            total_params += 1
-
-            if debug:
-                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
-            offset += unpartitioned_numel
-
-        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
-        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
-        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
-        # live optimizer object, so we are checking that the numbers are within the right range
-        align_to = 2 * world_size
-
-        def zero2_align(x):
-            return align_to * math.ceil(x / align_to)
-
-        if debug:
-            print(f"original offset={offset}, avail_numel={avail_numel}")
-
-        offset = zero2_align(offset)
-        avail_numel = zero2_align(avail_numel)
-
-        if debug:
-            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
-
-        # Sanity check
-        if offset != avail_numel:
-            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero2_merge_frozen_params(state_dict, zero_model_states)
-
-    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def zero3_partitioned_param_info(unpartitioned_numel, world_size):
-    remainder = unpartitioned_numel % world_size
-    padding_numel = (world_size - remainder) if remainder else 0
-    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
-    return partitioned_numel, padding_numel
-
-
-def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    if debug:
-        for i in range(world_size):
-            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
-            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in zero_model_states[0].frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
-        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-    avail_numel = fp32_flat_groups[0].numel() * world_size
-    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
-    # param, re-consolidating each param, while dealing with padding if any
-
-    # merge list of dicts, preserving order
-    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
-
-    if debug:
-        for i in range(world_size):
-            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
-
-        wanted_params = len(param_shapes)
-        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
-        # not asserting if there is a mismatch due to possible padding
-        avail_numel = fp32_flat_groups[0].numel() * world_size
-        print(f"Trainable params: Have {avail_numel} numels to process.")
-        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    offset = 0
-    total_numel = 0
-    total_params = 0
-    for name, shape in param_shapes.items():
-
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-        total_params += 1
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-        # XXX: memory usage doubles here
-        state_dict[name] = torch.cat(
-            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
-            0).narrow(0, 0, unpartitioned_numel).view(shape)
-        offset += partitioned_numel
-
-    offset *= world_size
-
-    # Sanity check
-    if offset != avail_numel:
-        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
-
-    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
-    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
-    via a model hub.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
-
-    Returns:
-        - pytorch ``state_dict``
-
-    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
-    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
-    the checkpoint.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
-        # do the training and checkpoint saving
-        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
-        model = model.cpu() # move to cpu
-        model.load_state_dict(state_dict)
-        # submit to model hub or save the model to share with others
-
-    In this example the ``model`` will no longer be usable in the deepspeed context of the same
-    application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
-
-    """
-    if tag is None:
-        latest_path = os.path.join(checkpoint_dir, 'latest')
-        if os.path.isfile(latest_path):
-            with open(latest_path, 'r') as fd:
-                tag = fd.read().strip()
-        else:
-            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
-
-    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
-
-    if not os.path.isdir(ds_checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
-
-    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
-
-
-def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
-    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-    """
-
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-    print(f"Saving fp32 state dict to {output_file}")
-    torch.save(state_dict, output_file)
-
-
-def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
-    """
-    1. Put the provided model to cpu
-    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
-    3. Load it into the provided model
-
-    Args:
-        - ``model``: the model object to update
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-
-    Returns:
-        - ``model`: modified model
-
-    Make sure you have plenty of CPU memory available before you call this function. If you don't
-    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
-    conveniently placed for you in the checkpoint folder.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
-        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
-        # submit to model hub or save the model to share with others
-
-    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
-    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    """
-    logger.info(f"Extracting fp32 weights")
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-
-    logger.info(f"Overwriting model with fp32 weights")
-    model = model.cpu()
-    model.load_state_dict(state_dict, strict=False)
-
-    return model
-
-
-if __name__ == "__main__":
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument("checkpoint_dir",
-                        type=str,
-                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
-    parser.add_argument(
-        "output_file",
-        type=str,
-        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
-    parser.add_argument("-t",
-                        "--tag",
-                        type=str,
-                        default=None,
-                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
-    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
-    args = parser.parse_args()
-
-    debug = args.debug
-
-    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
diff --git a/checkpoint-250/config.json b/checkpoint-250/config.json
deleted file mode 100644
index 82bacd02f94ab4a4cfdb7b12a0484fac8f301916..0000000000000000000000000000000000000000
--- a/checkpoint-250/config.json
+++ /dev/null
@@ -1,25 +0,0 @@
-{
-  "_name_or_path": "mistralai/Mistral-7B-v0.1",
-  "architectures": [
-    "MistralForCausalLM"
-  ],
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "hidden_act": "silu",
-  "hidden_size": 4096,
-  "initializer_range": 0.02,
-  "intermediate_size": 14336,
-  "max_position_embeddings": 32768,
-  "model_type": "mistral",
-  "num_attention_heads": 32,
-  "num_hidden_layers": 32,
-  "num_key_value_heads": 8,
-  "rms_norm_eps": 1e-05,
-  "rope_theta": 10000.0,
-  "sliding_window": 4096,
-  "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
-  "transformers_version": "4.34.0.dev0",
-  "use_cache": false,
-  "vocab_size": 32002
-}
diff --git a/checkpoint-250/generation_config.json b/checkpoint-250/generation_config.json
deleted file mode 100644
index 2c5f418036a121b3fd432d1bf2b3c5c9daf59fab..0000000000000000000000000000000000000000
--- a/checkpoint-250/generation_config.json
+++ /dev/null
@@ -1,6 +0,0 @@
-{
-  "_from_model_config": true,
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "transformers_version": "4.34.0.dev0"
-}
diff --git a/checkpoint-250/latest b/checkpoint-250/latest
deleted file mode 100644
index 87449ff1a854ba4a77ea33fbc24adaed3311d6b1..0000000000000000000000000000000000000000
--- a/checkpoint-250/latest
+++ /dev/null
@@ -1 +0,0 @@
-global_step250
\ No newline at end of file
diff --git a/checkpoint-250/pytorch_model-00001-of-00002.bin b/checkpoint-250/pytorch_model-00001-of-00002.bin
deleted file mode 100644
index c5b88b0093e91362eb3d554490f9d4e7d9c078e1..0000000000000000000000000000000000000000
--- a/checkpoint-250/pytorch_model-00001-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:0b50d9439b999b9fd5b6c2e695a1483624a4c6c6bcee46f35f79806dde564275
-size 9943044428
diff --git a/checkpoint-250/pytorch_model-00002-of-00002.bin b/checkpoint-250/pytorch_model-00002-of-00002.bin
deleted file mode 100644
index d82cccbceb6c3792ef28e09867f6b37bb9767afd..0000000000000000000000000000000000000000
--- a/checkpoint-250/pytorch_model-00002-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2a838f1a551e9ca10152f3cdb899a08d47a341ddb20b01839dc38a3eb6dac268
-size 4540552031
diff --git a/checkpoint-250/pytorch_model.bin.index.json b/checkpoint-250/pytorch_model.bin.index.json
deleted file mode 100644
index 53213fb82ddc02718be2ce686f00ba7fb0af95e7..0000000000000000000000000000000000000000
--- a/checkpoint-250/pytorch_model.bin.index.json
+++ /dev/null
@@ -1,298 +0,0 @@
-{
-  "metadata": {
-    "total_size": 14483496960
-  },
-  "weight_map": {
-    "lm_head.weight": "pytorch_model-00002-of-00002.bin",
-    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.23.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.norm.weight": "pytorch_model-00002-of-00002.bin"
-  }
-}
diff --git a/checkpoint-250/rng_state_0.pth b/checkpoint-250/rng_state_0.pth
deleted file mode 100644
index 24aba7bcc2a9fb783bd13a6a67b96e5ad055d89d..0000000000000000000000000000000000000000
--- a/checkpoint-250/rng_state_0.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1eafe3d5e0585dde8c5033613de99a5d4f23df4284a488f4007b3944580c0b97
-size 17655
diff --git a/checkpoint-250/rng_state_1.pth b/checkpoint-250/rng_state_1.pth
deleted file mode 100644
index 6b2ef88173fde17f2b3e738a28446f89a0528a96..0000000000000000000000000000000000000000
--- a/checkpoint-250/rng_state_1.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7e34eb456d2d003a2839f2daa9425e99bdd79ed7e24a1de9fc7d5738476bfb4b
-size 17655
diff --git a/checkpoint-250/rng_state_2.pth b/checkpoint-250/rng_state_2.pth
deleted file mode 100644
index 7b118d52a3006aea6c44f23f94c5568d1fb0a2f3..0000000000000000000000000000000000000000
--- a/checkpoint-250/rng_state_2.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b374af4a2765d8771cee7a72921d3c2e438b9bee34f0b2d098ce6071afeb65e4
-size 17655
diff --git a/checkpoint-250/rng_state_3.pth b/checkpoint-250/rng_state_3.pth
deleted file mode 100644
index 3f6fd9aa58eb1d5815ca991134531a3280601900..0000000000000000000000000000000000000000
--- a/checkpoint-250/rng_state_3.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5df75d8477fcc69c7abb03025313915ebfe3ac18c54a7c57aaa455c0099e13e5
-size 17655
diff --git a/checkpoint-250/trainer_state.json b/checkpoint-250/trainer_state.json
deleted file mode 100644
index cff5531f57c038af15c75c5f96c3d4bbe83b34d8..0000000000000000000000000000000000000000
--- a/checkpoint-250/trainer_state.json
+++ /dev/null
@@ -1,1543 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.03307753373908441,
-  "eval_steps": 756,
-  "global_step": 250,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 0.9197,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_loss": 1.4652303457260132,
-      "eval_runtime": 2.1726,
-      "eval_samples_per_second": 79.627,
-      "eval_steps_per_second": 3.682,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_bench_accuracy_agieval": 0.2711864406779661,
-      "eval_bench_accuracy_arc_challenge": 0.8703703703703703,
-      "eval_bench_accuracy_arc_easy": 0.9259259259259259,
-      "eval_bench_accuracy_bigbench": 0.36065573770491804,
-      "eval_bench_accuracy_boolq": 0.5740740740740741,
-      "eval_bench_accuracy_mmlu": 0.5185185185185185,
-      "eval_bench_accuracy_openbookqa": 0.1111111111111111,
-      "eval_bench_accuracy_truthful_qa": 0.3584905660377358,
-      "eval_bench_accuracy_winogrande": 0.4444444444444444,
-      "eval_bench_average_accuracy": 0.4927530209850072,
-      "eval_bench_loss": 2.6978388407144203,
-      "eval_bench_total_accuracy": 0.48893360160965793,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 1.3426,
-      "step": 2
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 1.5882,
-      "step": 3
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.8e-06,
-      "loss": 0.8542,
-      "step": 4
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 0.9629,
-      "step": 5
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3e-06,
-      "loss": 0.903,
-      "step": 6
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.6e-06,
-      "loss": 0.909,
-      "step": 7
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.2e-06,
-      "loss": 0.8666,
-      "step": 8
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 1.0108,
-      "step": 9
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.4e-06,
-      "loss": 0.8958,
-      "step": 10
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6e-06,
-      "loss": 0.9348,
-      "step": 11
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999602806831722e-06,
-      "loss": 0.7832,
-      "step": 12
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999205613663445e-06,
-      "loss": 0.8083,
-      "step": 13
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9988084204951675e-06,
-      "loss": 0.8164,
-      "step": 14
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99841122732689e-06,
-      "loss": 0.7834,
-      "step": 15
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.998014034158613e-06,
-      "loss": 0.8718,
-      "step": 16
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997616840990336e-06,
-      "loss": 0.84,
-      "step": 17
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997219647822058e-06,
-      "loss": 0.7397,
-      "step": 18
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99682245465378e-06,
-      "loss": 0.7445,
-      "step": 19
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996425261485502e-06,
-      "loss": 0.7898,
-      "step": 20
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996028068317225e-06,
-      "loss": 0.7388,
-      "step": 21
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9956308751489475e-06,
-      "loss": 0.7296,
-      "step": 22
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99523368198067e-06,
-      "loss": 0.7993,
-      "step": 23
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994836488812393e-06,
-      "loss": 0.7188,
-      "step": 24
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994439295644115e-06,
-      "loss": 0.7473,
-      "step": 25
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994042102475838e-06,
-      "loss": 0.6997,
-      "step": 26
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99364490930756e-06,
-      "loss": 0.725,
-      "step": 27
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.993247716139283e-06,
-      "loss": 0.7272,
-      "step": 28
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992850522971005e-06,
-      "loss": 0.7427,
-      "step": 29
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992453329802727e-06,
-      "loss": 0.7309,
-      "step": 30
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99205613663445e-06,
-      "loss": 0.6764,
-      "step": 31
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991658943466173e-06,
-      "loss": 0.7556,
-      "step": 32
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991261750297895e-06,
-      "loss": 0.7301,
-      "step": 33
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990864557129617e-06,
-      "loss": 0.6776,
-      "step": 34
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99046736396134e-06,
-      "loss": 0.6884,
-      "step": 35
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990070170793063e-06,
-      "loss": 0.7179,
-      "step": 36
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.989672977624785e-06,
-      "loss": 0.6915,
-      "step": 37
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.989275784456507e-06,
-      "loss": 0.7308,
-      "step": 38
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98887859128823e-06,
-      "loss": 0.6743,
-      "step": 39
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9884813981199526e-06,
-      "loss": 0.6604,
-      "step": 40
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.988084204951675e-06,
-      "loss": 0.6609,
-      "step": 41
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.987687011783397e-06,
-      "loss": 0.6524,
-      "step": 42
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98728981861512e-06,
-      "loss": 0.6386,
-      "step": 43
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986892625446843e-06,
-      "loss": 0.728,
-      "step": 44
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986495432278565e-06,
-      "loss": 0.6971,
-      "step": 45
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986098239110287e-06,
-      "loss": 0.6772,
-      "step": 46
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98570104594201e-06,
-      "loss": 0.6774,
-      "step": 47
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9853038527737325e-06,
-      "loss": 0.6868,
-      "step": 48
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984906659605455e-06,
-      "loss": 0.7169,
-      "step": 49
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984509466437178e-06,
-      "loss": 0.669,
-      "step": 50
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9841122732689e-06,
-      "loss": 0.7112,
-      "step": 51
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.983715080100622e-06,
-      "loss": 0.6667,
-      "step": 52
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.983317886932344e-06,
-      "loss": 0.6528,
-      "step": 53
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.982920693764068e-06,
-      "loss": 0.6699,
-      "step": 54
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98252350059579e-06,
-      "loss": 0.6584,
-      "step": 55
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9821263074275125e-06,
-      "loss": 0.6328,
-      "step": 56
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.981729114259235e-06,
-      "loss": 0.6472,
-      "step": 57
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.981331921090958e-06,
-      "loss": 0.6992,
-      "step": 58
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98093472792268e-06,
-      "loss": 0.6666,
-      "step": 59
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.980537534754402e-06,
-      "loss": 0.6819,
-      "step": 60
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.980140341586125e-06,
-      "loss": 0.705,
-      "step": 61
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.979743148417847e-06,
-      "loss": 0.6871,
-      "step": 62
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97934595524957e-06,
-      "loss": 0.6998,
-      "step": 63
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.978948762081292e-06,
-      "loss": 0.6081,
-      "step": 64
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9785515689130154e-06,
-      "loss": 0.6985,
-      "step": 65
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.978154375744738e-06,
-      "loss": 0.6631,
-      "step": 66
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97775718257646e-06,
-      "loss": 0.6534,
-      "step": 67
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.977359989408182e-06,
-      "loss": 0.6685,
-      "step": 68
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976962796239905e-06,
-      "loss": 0.6821,
-      "step": 69
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976565603071627e-06,
-      "loss": 0.6241,
-      "step": 70
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.976168409903349e-06,
-      "loss": 0.6357,
-      "step": 71
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.975771216735072e-06,
-      "loss": 0.6466,
-      "step": 72
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.975374023566795e-06,
-      "loss": 0.6579,
-      "step": 73
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9749768303985176e-06,
-      "loss": 0.6298,
-      "step": 74
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97457963723024e-06,
-      "loss": 0.703,
-      "step": 75
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.974182444061963e-06,
-      "loss": 0.6152,
-      "step": 76
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.973785250893685e-06,
-      "loss": 0.6682,
-      "step": 77
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.973388057725407e-06,
-      "loss": 0.6427,
-      "step": 78
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.972990864557129e-06,
-      "loss": 0.6969,
-      "step": 79
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.972593671388852e-06,
-      "loss": 0.6619,
-      "step": 80
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9721964782205745e-06,
-      "loss": 0.6332,
-      "step": 81
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9717992850522975e-06,
-      "loss": 0.6203,
-      "step": 82
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.97140209188402e-06,
-      "loss": 0.6463,
-      "step": 83
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.971004898715743e-06,
-      "loss": 0.6718,
-      "step": 84
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.970607705547465e-06,
-      "loss": 0.6495,
-      "step": 85
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.970210512379187e-06,
-      "loss": 0.5787,
-      "step": 86
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96981331921091e-06,
-      "loss": 0.6897,
-      "step": 87
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.969416126042632e-06,
-      "loss": 0.6688,
-      "step": 88
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9690189328743544e-06,
-      "loss": 0.6697,
-      "step": 89
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.968621739706077e-06,
-      "loss": 0.6156,
-      "step": 90
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9682245465378e-06,
-      "loss": 0.6301,
-      "step": 91
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967827353369523e-06,
-      "loss": 0.6121,
-      "step": 92
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967430160201245e-06,
-      "loss": 0.6177,
-      "step": 93
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.967032967032967e-06,
-      "loss": 0.611,
-      "step": 94
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96663577386469e-06,
-      "loss": 0.6359,
-      "step": 95
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.966238580696412e-06,
-      "loss": 0.6417,
-      "step": 96
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.965841387528134e-06,
-      "loss": 0.6312,
-      "step": 97
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.965444194359857e-06,
-      "loss": 0.6184,
-      "step": 98
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9650470011915796e-06,
-      "loss": 0.6724,
-      "step": 99
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.964649808023302e-06,
-      "loss": 0.6833,
-      "step": 100
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.964252614855025e-06,
-      "loss": 0.6433,
-      "step": 101
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.963855421686747e-06,
-      "loss": 0.6766,
-      "step": 102
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96345822851847e-06,
-      "loss": 0.6527,
-      "step": 103
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.963061035350192e-06,
-      "loss": 0.5982,
-      "step": 104
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.962663842181914e-06,
-      "loss": 0.6749,
-      "step": 105
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.962266649013637e-06,
-      "loss": 0.6494,
-      "step": 106
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9618694558453595e-06,
-      "loss": 0.6998,
-      "step": 107
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.961472262677082e-06,
-      "loss": 0.6112,
-      "step": 108
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.961075069508805e-06,
-      "loss": 0.624,
-      "step": 109
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.960677876340528e-06,
-      "loss": 0.6329,
-      "step": 110
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.96028068317225e-06,
-      "loss": 0.6491,
-      "step": 111
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.959883490003972e-06,
-      "loss": 0.6672,
-      "step": 112
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.959486296835694e-06,
-      "loss": 0.6279,
-      "step": 113
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.959089103667417e-06,
-      "loss": 0.6479,
-      "step": 114
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9586919104991395e-06,
-      "loss": 0.6214,
-      "step": 115
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.958294717330862e-06,
-      "loss": 0.6618,
-      "step": 116
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957897524162585e-06,
-      "loss": 0.6703,
-      "step": 117
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957500330994307e-06,
-      "loss": 0.6417,
-      "step": 118
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.957103137826029e-06,
-      "loss": 0.631,
-      "step": 119
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.956705944657752e-06,
-      "loss": 0.6169,
-      "step": 120
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.956308751489475e-06,
-      "loss": 0.6521,
-      "step": 121
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955911558321197e-06,
-      "loss": 0.6635,
-      "step": 122
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955514365152919e-06,
-      "loss": 0.6496,
-      "step": 123
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.955117171984642e-06,
-      "loss": 0.6431,
-      "step": 124
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.954719978816365e-06,
-      "loss": 0.6246,
-      "step": 125
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.954322785648087e-06,
-      "loss": 0.6557,
-      "step": 126
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953925592479809e-06,
-      "loss": 0.6082,
-      "step": 127
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953528399311532e-06,
-      "loss": 0.5941,
-      "step": 128
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.953131206143255e-06,
-      "loss": 0.6566,
-      "step": 129
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.952734012974977e-06,
-      "loss": 0.6243,
-      "step": 130
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.952336819806699e-06,
-      "loss": 0.594,
-      "step": 131
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.951939626638422e-06,
-      "loss": 0.68,
-      "step": 132
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9515424334701446e-06,
-      "loss": 0.6302,
-      "step": 133
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.951145240301867e-06,
-      "loss": 0.6251,
-      "step": 134
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.950748047133589e-06,
-      "loss": 0.6326,
-      "step": 135
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.950350853965312e-06,
-      "loss": 0.6314,
-      "step": 136
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949953660797034e-06,
-      "loss": 0.6598,
-      "step": 137
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949556467628757e-06,
-      "loss": 0.6583,
-      "step": 138
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.949159274460479e-06,
-      "loss": 0.6162,
-      "step": 139
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.948762081292202e-06,
-      "loss": 0.7042,
-      "step": 140
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9483648881239245e-06,
-      "loss": 0.6733,
-      "step": 141
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.947967694955647e-06,
-      "loss": 0.6103,
-      "step": 142
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94757050178737e-06,
-      "loss": 0.6269,
-      "step": 143
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.947173308619092e-06,
-      "loss": 0.663,
-      "step": 144
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.946776115450814e-06,
-      "loss": 0.5794,
-      "step": 145
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.946378922282537e-06,
-      "loss": 0.6868,
-      "step": 146
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.945981729114259e-06,
-      "loss": 0.6064,
-      "step": 147
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.945584535945982e-06,
-      "loss": 0.6519,
-      "step": 148
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9451873427777045e-06,
-      "loss": 0.655,
-      "step": 149
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.944790149609427e-06,
-      "loss": 0.6617,
-      "step": 150
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94439295644115e-06,
-      "loss": 0.627,
-      "step": 151
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.943995763272872e-06,
-      "loss": 0.5837,
-      "step": 152
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.943598570104594e-06,
-      "loss": 0.6201,
-      "step": 153
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.943201376936317e-06,
-      "loss": 0.6291,
-      "step": 154
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.942804183768039e-06,
-      "loss": 0.6061,
-      "step": 155
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.942406990599761e-06,
-      "loss": 0.624,
-      "step": 156
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.942009797431484e-06,
-      "loss": 0.6418,
-      "step": 157
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.941612604263207e-06,
-      "loss": 0.5858,
-      "step": 158
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.94121541109493e-06,
-      "loss": 0.6407,
-      "step": 159
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.940818217926652e-06,
-      "loss": 0.6222,
-      "step": 160
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.940421024758374e-06,
-      "loss": 0.5938,
-      "step": 161
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.940023831590097e-06,
-      "loss": 0.6157,
-      "step": 162
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.939626638421819e-06,
-      "loss": 0.5989,
-      "step": 163
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.939229445253541e-06,
-      "loss": 0.7056,
-      "step": 164
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.938832252085264e-06,
-      "loss": 0.6606,
-      "step": 165
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9384350589169865e-06,
-      "loss": 0.6303,
-      "step": 166
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9380378657487095e-06,
-      "loss": 0.6332,
-      "step": 167
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.937640672580432e-06,
-      "loss": 0.6197,
-      "step": 168
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.937243479412155e-06,
-      "loss": 0.6318,
-      "step": 169
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.936846286243877e-06,
-      "loss": 0.6598,
-      "step": 170
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.936449093075599e-06,
-      "loss": 0.662,
-      "step": 171
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.936051899907321e-06,
-      "loss": 0.6018,
-      "step": 172
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.935654706739044e-06,
-      "loss": 0.6955,
-      "step": 173
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.9352575135707665e-06,
-      "loss": 0.6283,
-      "step": 174
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.934860320402489e-06,
-      "loss": 0.6829,
-      "step": 175
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.934463127234212e-06,
-      "loss": 0.5985,
-      "step": 176
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.934065934065935e-06,
-      "loss": 0.6385,
-      "step": 177
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.933668740897657e-06,
-      "loss": 0.6326,
-      "step": 178
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.933271547729379e-06,
-      "loss": 0.639,
-      "step": 179
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.932874354561102e-06,
-      "loss": 0.6084,
-      "step": 180
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.932477161392824e-06,
-      "loss": 0.6549,
-      "step": 181
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.932079968224546e-06,
-      "loss": 0.6728,
-      "step": 182
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.931682775056269e-06,
-      "loss": 0.6351,
-      "step": 183
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.931285581887992e-06,
-      "loss": 0.6375,
-      "step": 184
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.930888388719714e-06,
-      "loss": 0.6814,
-      "step": 185
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.930491195551437e-06,
-      "loss": 0.5968,
-      "step": 186
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.930094002383159e-06,
-      "loss": 0.6053,
-      "step": 187
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.929696809214882e-06,
-      "loss": 0.6468,
-      "step": 188
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.929299616046604e-06,
-      "loss": 0.6407,
-      "step": 189
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.928902422878326e-06,
-      "loss": 0.6996,
-      "step": 190
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.928505229710049e-06,
-      "loss": 0.6158,
-      "step": 191
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9281080365417716e-06,
-      "loss": 0.6128,
-      "step": 192
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.927710843373494e-06,
-      "loss": 0.6558,
-      "step": 193
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.927313650205216e-06,
-      "loss": 0.6726,
-      "step": 194
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.92691645703694e-06,
-      "loss": 0.6292,
-      "step": 195
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.926519263868662e-06,
-      "loss": 0.6004,
-      "step": 196
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.926122070700384e-06,
-      "loss": 0.599,
-      "step": 197
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.925724877532106e-06,
-      "loss": 0.6374,
-      "step": 198
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.925327684363829e-06,
-      "loss": 0.6472,
-      "step": 199
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9249304911955515e-06,
-      "loss": 0.594,
-      "step": 200
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.924533298027274e-06,
-      "loss": 0.6382,
-      "step": 201
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.924136104858997e-06,
-      "loss": 0.5817,
-      "step": 202
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.923738911690719e-06,
-      "loss": 0.6128,
-      "step": 203
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.923341718522442e-06,
-      "loss": 0.651,
-      "step": 204
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.922944525354164e-06,
-      "loss": 0.5681,
-      "step": 205
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.922547332185887e-06,
-      "loss": 0.6183,
-      "step": 206
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.922150139017609e-06,
-      "loss": 0.5867,
-      "step": 207
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9217529458493315e-06,
-      "loss": 0.6048,
-      "step": 208
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.921355752681054e-06,
-      "loss": 0.6968,
-      "step": 209
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.920958559512777e-06,
-      "loss": 0.6259,
-      "step": 210
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.920561366344499e-06,
-      "loss": 0.6076,
-      "step": 211
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.920164173176221e-06,
-      "loss": 0.64,
-      "step": 212
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.919766980007944e-06,
-      "loss": 0.6249,
-      "step": 213
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.919369786839667e-06,
-      "loss": 0.6331,
-      "step": 214
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.918972593671389e-06,
-      "loss": 0.6466,
-      "step": 215
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.918575400503111e-06,
-      "loss": 0.5982,
-      "step": 216
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9181782073348344e-06,
-      "loss": 0.5719,
-      "step": 217
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.917781014166557e-06,
-      "loss": 0.6032,
-      "step": 218
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.917383820998279e-06,
-      "loss": 0.5741,
-      "step": 219
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.916986627830001e-06,
-      "loss": 0.58,
-      "step": 220
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.916589434661724e-06,
-      "loss": 0.6232,
-      "step": 221
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.916192241493446e-06,
-      "loss": 0.6287,
-      "step": 222
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.915795048325169e-06,
-      "loss": 0.6344,
-      "step": 223
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.915397855156891e-06,
-      "loss": 0.6536,
-      "step": 224
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.915000661988614e-06,
-      "loss": 0.6297,
-      "step": 225
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9146034688203365e-06,
-      "loss": 0.5635,
-      "step": 226
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.914206275652059e-06,
-      "loss": 0.5931,
-      "step": 227
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.913809082483782e-06,
-      "loss": 0.5681,
-      "step": 228
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.913411889315504e-06,
-      "loss": 0.6155,
-      "step": 229
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.913014696147226e-06,
-      "loss": 0.605,
-      "step": 230
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.912617502978948e-06,
-      "loss": 0.6364,
-      "step": 231
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.912220309810671e-06,
-      "loss": 0.6333,
-      "step": 232
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.911823116642394e-06,
-      "loss": 0.6666,
-      "step": 233
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9114259234741165e-06,
-      "loss": 0.6296,
-      "step": 234
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.911028730305839e-06,
-      "loss": 0.6422,
-      "step": 235
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.910631537137562e-06,
-      "loss": 0.6426,
-      "step": 236
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.910234343969284e-06,
-      "loss": 0.6389,
-      "step": 237
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.909837150801006e-06,
-      "loss": 0.5695,
-      "step": 238
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.909439957632729e-06,
-      "loss": 0.6271,
-      "step": 239
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.909042764464451e-06,
-      "loss": 0.5981,
-      "step": 240
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.908645571296173e-06,
-      "loss": 0.6345,
-      "step": 241
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9082483781278964e-06,
-      "loss": 0.6404,
-      "step": 242
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.9078511849596195e-06,
-      "loss": 0.6046,
-      "step": 243
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.907453991791342e-06,
-      "loss": 0.5791,
-      "step": 244
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.907056798623064e-06,
-      "loss": 0.623,
-      "step": 245
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.906659605454786e-06,
-      "loss": 0.6274,
-      "step": 246
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.906262412286509e-06,
-      "loss": 0.629,
-      "step": 247
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.905865219118231e-06,
-      "loss": 0.6967,
-      "step": 248
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.905468025949953e-06,
-      "loss": 0.6649,
-      "step": 249
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 5.905070832781676e-06,
-      "loss": 0.6514,
-      "step": 250
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 15116,
-  "num_train_epochs": 2,
-  "save_steps": 50,
-  "total_flos": 1.048510703075328e+18,
-  "trial_name": null,
-  "trial_params": null
-}
diff --git a/checkpoint-250/training_args.bin b/checkpoint-250/training_args.bin
deleted file mode 100644
index 731dabe55350d521ab0dfde0b2f023771c347250..0000000000000000000000000000000000000000
--- a/checkpoint-250/training_args.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2f05be88d930176935da1678b48a8294634889bf7ae4f8bebdbaca140c2dac08
-size 5947
diff --git a/checkpoint-250/zero_to_fp32.py b/checkpoint-250/zero_to_fp32.py
deleted file mode 100755
index c98caae31534368be22b67fc4ae906836c992a8d..0000000000000000000000000000000000000000
--- a/checkpoint-250/zero_to_fp32.py
+++ /dev/null
@@ -1,587 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright (c) Microsoft Corporation.
-# SPDX-License-Identifier: Apache-2.0
-
-# DeepSpeed Team
-
-# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
-# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
-# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
-# application.
-#
-# example: python zero_to_fp32.py . pytorch_model.bin
-
-import argparse
-import torch
-import glob
-import math
-import os
-import re
-from collections import OrderedDict
-from dataclasses import dataclass
-
-# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
-# DeepSpeed data structures it has to be available in the current python environment.
-from deepspeed.utils import logger
-from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
-                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
-                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
-
-
-@dataclass
-class zero_model_state:
-    buffers: dict()
-    param_shapes: dict()
-    shared_params: list
-    ds_version: int
-    frozen_param_shapes: dict()
-    frozen_param_fragments: dict()
-
-
-debug = 0
-
-# load to cpu
-device = torch.device('cpu')
-
-
-def atoi(text):
-    return int(text) if text.isdigit() else text
-
-
-def natural_keys(text):
-    '''
-    alist.sort(key=natural_keys) sorts in human order
-    http://nedbatchelder.com/blog/200712/human_sorting.html
-    (See Toothy's implementation in the comments)
-    '''
-    return [atoi(c) for c in re.split(r'(\d+)', text)]
-
-
-def get_model_state_file(checkpoint_dir, zero_stage):
-    if not os.path.isdir(checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
-
-    # there should be only one file
-    if zero_stage <= 2:
-        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
-    elif zero_stage == 3:
-        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
-
-    if not os.path.exists(file):
-        raise FileNotFoundError(f"can't find model states file at '{file}'")
-
-    return file
-
-
-def get_checkpoint_files(checkpoint_dir, glob_pattern):
-    # XXX: need to test that this simple glob rule works for multi-node setup too
-    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
-
-    if len(ckpt_files) == 0:
-        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
-
-    return ckpt_files
-
-
-def get_optim_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
-
-
-def get_model_state_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
-
-
-def parse_model_states(files):
-    zero_model_states = []
-    for file in files:
-        state_dict = torch.load(file, map_location=device)
-
-        if BUFFER_NAMES not in state_dict:
-            raise ValueError(f"{file} is not a model state checkpoint")
-        buffer_names = state_dict[BUFFER_NAMES]
-        if debug:
-            print("Found buffers:", buffer_names)
-
-        # recover just the buffers while restoring them to fp32 if they were saved in fp16
-        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
-        param_shapes = state_dict[PARAM_SHAPES]
-
-        # collect parameters that are included in param_shapes
-        param_names = []
-        for s in param_shapes:
-            for name in s.keys():
-                param_names.append(name)
-
-        # update with frozen parameters
-        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
-        if frozen_param_shapes is not None:
-            if debug:
-                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
-            param_names += list(frozen_param_shapes.keys())
-
-        # handle shared params
-        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
-
-        ds_version = state_dict.get(DS_VERSION, None)
-
-        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
-
-        z_model_state = zero_model_state(buffers=buffers,
-                                         param_shapes=param_shapes,
-                                         shared_params=shared_params,
-                                         ds_version=ds_version,
-                                         frozen_param_shapes=frozen_param_shapes,
-                                         frozen_param_fragments=frozen_param_fragments)
-        zero_model_states.append(z_model_state)
-
-    return zero_model_states
-
-
-def parse_optim_states(files, ds_checkpoint_dir):
-
-    total_files = len(files)
-    state_dicts = []
-    for f in files:
-        state_dict = torch.load(f, map_location=device)
-        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
-        # and also handle the case where it was already removed by another helper script
-        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
-        state_dicts.append(state_dict)
-
-    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
-        raise ValueError(f"{files[0]} is not a zero checkpoint")
-    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
-    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
-
-    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
-    # parameters can be different from data parallelism for non-expert parameters. So we can just
-    # use the max of the partition_count to get the dp world_size.
-
-    if type(world_size) is list:
-        world_size = max(world_size)
-
-    if world_size != total_files:
-        raise ValueError(
-            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
-            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
-        )
-
-    # the groups are named differently in each stage
-    if zero_stage <= 2:
-        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
-    elif zero_stage == 3:
-        fp32_groups_key = FP32_FLAT_GROUPS
-    else:
-        raise ValueError(f"unknown zero stage {zero_stage}")
-
-    if zero_stage <= 2:
-        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
-    elif zero_stage == 3:
-        # if there is more than one param group, there will be multiple flattened tensors - one
-        # flattened tensor per group - for simplicity merge them into a single tensor
-        #
-        # XXX: could make the script more memory efficient for when there are multiple groups - it
-        # will require matching the sub-lists of param_shapes for each param group flattened tensor
-
-        fp32_flat_groups = [
-            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
-        ]
-
-    return zero_stage, world_size, fp32_flat_groups
-
-
-def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
-    """
-    Returns fp32 state_dict reconstructed from ds checkpoint
-
-    Args:
-        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
-
-    """
-    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
-
-    optim_files = get_optim_files(ds_checkpoint_dir)
-    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
-    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
-
-    model_files = get_model_state_files(ds_checkpoint_dir)
-
-    zero_model_states = parse_model_states(model_files)
-    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
-
-    if zero_stage <= 2:
-        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-    elif zero_stage == 3:
-        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-
-
-def _zero2_merge_frozen_params(state_dict, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
-
-    if debug:
-        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
-        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        state_dict[name] = frozen_param_fragments[name]
-
-        if debug:
-            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-
-    # Reconstruction protocol:
-    #
-    # XXX: document this
-
-    if debug:
-        for i in range(world_size):
-            for j in range(len(fp32_flat_groups[0])):
-                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
-
-    # XXX: memory usage doubles here (zero2)
-    num_param_groups = len(fp32_flat_groups[0])
-    merged_single_partition_of_fp32_groups = []
-    for i in range(num_param_groups):
-        merged_partitions = [sd[i] for sd in fp32_flat_groups]
-        full_single_fp32_vector = torch.cat(merged_partitions, 0)
-        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
-    avail_numel = sum(
-        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
-
-    if debug:
-        wanted_params = sum([len(shapes) for shapes in param_shapes])
-        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
-        # not asserting if there is a mismatch due to possible padding
-        print(f"Have {avail_numel} numels to process.")
-        print(f"Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    total_numel = 0
-    total_params = 0
-    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
-        offset = 0
-        avail_numel = full_single_fp32_vector.numel()
-        for name, shape in shapes.items():
-
-            unpartitioned_numel = shape.numel()
-            total_numel += unpartitioned_numel
-            total_params += 1
-
-            if debug:
-                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
-            offset += unpartitioned_numel
-
-        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
-        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
-        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
-        # live optimizer object, so we are checking that the numbers are within the right range
-        align_to = 2 * world_size
-
-        def zero2_align(x):
-            return align_to * math.ceil(x / align_to)
-
-        if debug:
-            print(f"original offset={offset}, avail_numel={avail_numel}")
-
-        offset = zero2_align(offset)
-        avail_numel = zero2_align(avail_numel)
-
-        if debug:
-            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
-
-        # Sanity check
-        if offset != avail_numel:
-            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero2_merge_frozen_params(state_dict, zero_model_states)
-
-    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def zero3_partitioned_param_info(unpartitioned_numel, world_size):
-    remainder = unpartitioned_numel % world_size
-    padding_numel = (world_size - remainder) if remainder else 0
-    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
-    return partitioned_numel, padding_numel
-
-
-def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    if debug:
-        for i in range(world_size):
-            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
-            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in zero_model_states[0].frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
-        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-    avail_numel = fp32_flat_groups[0].numel() * world_size
-    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
-    # param, re-consolidating each param, while dealing with padding if any
-
-    # merge list of dicts, preserving order
-    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
-
-    if debug:
-        for i in range(world_size):
-            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
-
-        wanted_params = len(param_shapes)
-        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
-        # not asserting if there is a mismatch due to possible padding
-        avail_numel = fp32_flat_groups[0].numel() * world_size
-        print(f"Trainable params: Have {avail_numel} numels to process.")
-        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    offset = 0
-    total_numel = 0
-    total_params = 0
-    for name, shape in param_shapes.items():
-
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-        total_params += 1
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-        # XXX: memory usage doubles here
-        state_dict[name] = torch.cat(
-            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
-            0).narrow(0, 0, unpartitioned_numel).view(shape)
-        offset += partitioned_numel
-
-    offset *= world_size
-
-    # Sanity check
-    if offset != avail_numel:
-        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
-
-    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
-    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
-    via a model hub.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
-
-    Returns:
-        - pytorch ``state_dict``
-
-    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
-    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
-    the checkpoint.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
-        # do the training and checkpoint saving
-        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
-        model = model.cpu() # move to cpu
-        model.load_state_dict(state_dict)
-        # submit to model hub or save the model to share with others
-
-    In this example the ``model`` will no longer be usable in the deepspeed context of the same
-    application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
-
-    """
-    if tag is None:
-        latest_path = os.path.join(checkpoint_dir, 'latest')
-        if os.path.isfile(latest_path):
-            with open(latest_path, 'r') as fd:
-                tag = fd.read().strip()
-        else:
-            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
-
-    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
-
-    if not os.path.isdir(ds_checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
-
-    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
-
-
-def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
-    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-    """
-
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-    print(f"Saving fp32 state dict to {output_file}")
-    torch.save(state_dict, output_file)
-
-
-def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
-    """
-    1. Put the provided model to cpu
-    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
-    3. Load it into the provided model
-
-    Args:
-        - ``model``: the model object to update
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-
-    Returns:
-        - ``model`: modified model
-
-    Make sure you have plenty of CPU memory available before you call this function. If you don't
-    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
-    conveniently placed for you in the checkpoint folder.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
-        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
-        # submit to model hub or save the model to share with others
-
-    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
-    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    """
-    logger.info(f"Extracting fp32 weights")
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-
-    logger.info(f"Overwriting model with fp32 weights")
-    model = model.cpu()
-    model.load_state_dict(state_dict, strict=False)
-
-    return model
-
-
-if __name__ == "__main__":
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument("checkpoint_dir",
-                        type=str,
-                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
-    parser.add_argument(
-        "output_file",
-        type=str,
-        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
-    parser.add_argument("-t",
-                        "--tag",
-                        type=str,
-                        default=None,
-                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
-    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
-    args = parser.parse_args()
-
-    debug = args.debug
-
-    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
diff --git a/ckpt50/config.json b/ckpt50/config.json
deleted file mode 100644
index 82bacd02f94ab4a4cfdb7b12a0484fac8f301916..0000000000000000000000000000000000000000
--- a/ckpt50/config.json
+++ /dev/null
@@ -1,25 +0,0 @@
-{
-  "_name_or_path": "mistralai/Mistral-7B-v0.1",
-  "architectures": [
-    "MistralForCausalLM"
-  ],
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "hidden_act": "silu",
-  "hidden_size": 4096,
-  "initializer_range": 0.02,
-  "intermediate_size": 14336,
-  "max_position_embeddings": 32768,
-  "model_type": "mistral",
-  "num_attention_heads": 32,
-  "num_hidden_layers": 32,
-  "num_key_value_heads": 8,
-  "rms_norm_eps": 1e-05,
-  "rope_theta": 10000.0,
-  "sliding_window": 4096,
-  "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
-  "transformers_version": "4.34.0.dev0",
-  "use_cache": false,
-  "vocab_size": 32002
-}
diff --git a/ckpt50/generation_config.json b/ckpt50/generation_config.json
deleted file mode 100644
index 2c5f418036a121b3fd432d1bf2b3c5c9daf59fab..0000000000000000000000000000000000000000
--- a/ckpt50/generation_config.json
+++ /dev/null
@@ -1,6 +0,0 @@
-{
-  "_from_model_config": true,
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "transformers_version": "4.34.0.dev0"
-}
diff --git a/ckpt50/latest b/ckpt50/latest
deleted file mode 100644
index 9b4dc801e3fb152ef5c0ee60d309c705a9b01564..0000000000000000000000000000000000000000
--- a/ckpt50/latest
+++ /dev/null
@@ -1 +0,0 @@
-global_step50
\ No newline at end of file
diff --git a/ckpt50/pytorch_model-00001-of-00002.bin b/ckpt50/pytorch_model-00001-of-00002.bin
deleted file mode 100644
index 5763f40101dbcef8a5301e8e4f8143c5ee3a0214..0000000000000000000000000000000000000000
--- a/ckpt50/pytorch_model-00001-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:74a299c2c28cb9b40f392afd5819e34cdf713bb77c71aca3038602dab286095c
-size 9943044428
diff --git a/ckpt50/pytorch_model-00002-of-00002.bin b/ckpt50/pytorch_model-00002-of-00002.bin
deleted file mode 100644
index 69d5a838ffdef727c6be26e344738723ad7844c0..0000000000000000000000000000000000000000
--- a/ckpt50/pytorch_model-00002-of-00002.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:078f72ce86f271f6edfb8ca9e12be814398f4b4223c1ccfdc7415da377ff3e1f
-size 4540552031
diff --git a/ckpt50/pytorch_model.bin.index.json b/ckpt50/pytorch_model.bin.index.json
deleted file mode 100644
index 53213fb82ddc02718be2ce686f00ba7fb0af95e7..0000000000000000000000000000000000000000
--- a/ckpt50/pytorch_model.bin.index.json
+++ /dev/null
@@ -1,298 +0,0 @@
-{
-  "metadata": {
-    "total_size": 14483496960
-  },
-  "weight_map": {
-    "lm_head.weight": "pytorch_model-00002-of-00002.bin",
-    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.23.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
-    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
-    "model.norm.weight": "pytorch_model-00002-of-00002.bin"
-  }
-}
diff --git a/ckpt50/rng_state_0.pth b/ckpt50/rng_state_0.pth
deleted file mode 100644
index 24aba7bcc2a9fb783bd13a6a67b96e5ad055d89d..0000000000000000000000000000000000000000
--- a/ckpt50/rng_state_0.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1eafe3d5e0585dde8c5033613de99a5d4f23df4284a488f4007b3944580c0b97
-size 17655
diff --git a/ckpt50/rng_state_1.pth b/ckpt50/rng_state_1.pth
deleted file mode 100644
index 6b2ef88173fde17f2b3e738a28446f89a0528a96..0000000000000000000000000000000000000000
--- a/ckpt50/rng_state_1.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7e34eb456d2d003a2839f2daa9425e99bdd79ed7e24a1de9fc7d5738476bfb4b
-size 17655
diff --git a/ckpt50/rng_state_2.pth b/ckpt50/rng_state_2.pth
deleted file mode 100644
index 7b118d52a3006aea6c44f23f94c5568d1fb0a2f3..0000000000000000000000000000000000000000
--- a/ckpt50/rng_state_2.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b374af4a2765d8771cee7a72921d3c2e438b9bee34f0b2d098ce6071afeb65e4
-size 17655
diff --git a/ckpt50/rng_state_3.pth b/ckpt50/rng_state_3.pth
deleted file mode 100644
index 3f6fd9aa58eb1d5815ca991134531a3280601900..0000000000000000000000000000000000000000
--- a/ckpt50/rng_state_3.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5df75d8477fcc69c7abb03025313915ebfe3ac18c54a7c57aaa455c0099e13e5
-size 17655
diff --git a/ckpt50/trainer_state.json b/ckpt50/trainer_state.json
deleted file mode 100644
index 052244bb1c15d29f937923a54560092bc86574ed..0000000000000000000000000000000000000000
--- a/ckpt50/trainer_state.json
+++ /dev/null
@@ -1,343 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.006615506747816882,
-  "eval_steps": 756,
-  "global_step": 50,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 0.929,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_loss": 0.9224275946617126,
-      "eval_runtime": 2.17,
-      "eval_samples_per_second": 79.724,
-      "eval_steps_per_second": 3.687,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "eval_bench_accuracy_agieval": 0.288135593220339,
-      "eval_bench_accuracy_arc_challenge": 0.8148148148148148,
-      "eval_bench_accuracy_arc_easy": 0.9074074074074074,
-      "eval_bench_accuracy_bigbench": 0.3442622950819672,
-      "eval_bench_accuracy_boolq": 0.5185185185185185,
-      "eval_bench_accuracy_mmlu": 0.48148148148148145,
-      "eval_bench_accuracy_openbookqa": 0.14814814814814814,
-      "eval_bench_accuracy_truthful_qa": 0.37735849056603776,
-      "eval_bench_accuracy_winogrande": 0.4074074074074074,
-      "eval_bench_average_accuracy": 0.4763926840717912,
-      "eval_bench_loss": 5.786159653261484,
-      "eval_bench_total_accuracy": 0.47283702213279677,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 0.8533,
-      "step": 2
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 0.9641,
-      "step": 3
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.8e-06,
-      "loss": 0.8488,
-      "step": 4
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 0.8863,
-      "step": 5
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3e-06,
-      "loss": 0.7988,
-      "step": 6
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3.6e-06,
-      "loss": 0.7789,
-      "step": 7
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.2e-06,
-      "loss": 0.7144,
-      "step": 8
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 0.8322,
-      "step": 9
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.4e-06,
-      "loss": 0.734,
-      "step": 10
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6e-06,
-      "loss": 0.7861,
-      "step": 11
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999602806831722e-06,
-      "loss": 0.6733,
-      "step": 12
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.999205613663445e-06,
-      "loss": 0.7019,
-      "step": 13
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9988084204951675e-06,
-      "loss": 0.7096,
-      "step": 14
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99841122732689e-06,
-      "loss": 0.6745,
-      "step": 15
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.998014034158613e-06,
-      "loss": 0.8022,
-      "step": 16
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997616840990336e-06,
-      "loss": 0.7753,
-      "step": 17
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.997219647822058e-06,
-      "loss": 0.6939,
-      "step": 18
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99682245465378e-06,
-      "loss": 0.689,
-      "step": 19
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996425261485502e-06,
-      "loss": 0.7419,
-      "step": 20
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.996028068317225e-06,
-      "loss": 0.6975,
-      "step": 21
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.9956308751489475e-06,
-      "loss": 0.686,
-      "step": 22
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99523368198067e-06,
-      "loss": 0.7576,
-      "step": 23
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994836488812393e-06,
-      "loss": 0.6802,
-      "step": 24
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994439295644115e-06,
-      "loss": 0.711,
-      "step": 25
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.994042102475838e-06,
-      "loss": 0.6658,
-      "step": 26
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99364490930756e-06,
-      "loss": 0.685,
-      "step": 27
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.993247716139283e-06,
-      "loss": 0.6881,
-      "step": 28
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992850522971005e-06,
-      "loss": 0.7066,
-      "step": 29
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.992453329802727e-06,
-      "loss": 0.6993,
-      "step": 30
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99205613663445e-06,
-      "loss": 0.6429,
-      "step": 31
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991658943466173e-06,
-      "loss": 0.7205,
-      "step": 32
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.991261750297895e-06,
-      "loss": 0.703,
-      "step": 33
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990864557129617e-06,
-      "loss": 0.6512,
-      "step": 34
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.99046736396134e-06,
-      "loss": 0.6583,
-      "step": 35
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.990070170793063e-06,
-      "loss": 0.6904,
-      "step": 36
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.989672977624785e-06,
-      "loss": 0.6619,
-      "step": 37
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.989275784456507e-06,
-      "loss": 0.7033,
-      "step": 38
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98887859128823e-06,
-      "loss": 0.6522,
-      "step": 39
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9884813981199526e-06,
-      "loss": 0.6365,
-      "step": 40
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.988084204951675e-06,
-      "loss": 0.6422,
-      "step": 41
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.987687011783397e-06,
-      "loss": 0.625,
-      "step": 42
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98728981861512e-06,
-      "loss": 0.6148,
-      "step": 43
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986892625446843e-06,
-      "loss": 0.7089,
-      "step": 44
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986495432278565e-06,
-      "loss": 0.681,
-      "step": 45
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.986098239110287e-06,
-      "loss": 0.6505,
-      "step": 46
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.98570104594201e-06,
-      "loss": 0.6531,
-      "step": 47
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.9853038527737325e-06,
-      "loss": 0.6683,
-      "step": 48
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984906659605455e-06,
-      "loss": 0.6967,
-      "step": 49
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.984509466437178e-06,
-      "loss": 0.6514,
-      "step": 50
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 15116,
-  "num_train_epochs": 2,
-  "save_steps": 50,
-  "total_flos": 2.097021406150656e+17,
-  "trial_name": null,
-  "trial_params": null
-}
diff --git a/ckpt50/training_args.bin b/ckpt50/training_args.bin
deleted file mode 100644
index fd48cb2c1f189a745f6dd9ed1b31fb65735ab74f..0000000000000000000000000000000000000000
--- a/ckpt50/training_args.bin
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d199aaf2c0faf27119865363ae8807d2a26fe0d807759935113104fe056a8359
-size 5947
diff --git a/ckpt50/zero_to_fp32.py b/ckpt50/zero_to_fp32.py
deleted file mode 100755
index c98caae31534368be22b67fc4ae906836c992a8d..0000000000000000000000000000000000000000
--- a/ckpt50/zero_to_fp32.py
+++ /dev/null
@@ -1,587 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright (c) Microsoft Corporation.
-# SPDX-License-Identifier: Apache-2.0
-
-# DeepSpeed Team
-
-# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
-# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
-# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
-# application.
-#
-# example: python zero_to_fp32.py . pytorch_model.bin
-
-import argparse
-import torch
-import glob
-import math
-import os
-import re
-from collections import OrderedDict
-from dataclasses import dataclass
-
-# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
-# DeepSpeed data structures it has to be available in the current python environment.
-from deepspeed.utils import logger
-from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
-                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
-                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
-
-
-@dataclass
-class zero_model_state:
-    buffers: dict()
-    param_shapes: dict()
-    shared_params: list
-    ds_version: int
-    frozen_param_shapes: dict()
-    frozen_param_fragments: dict()
-
-
-debug = 0
-
-# load to cpu
-device = torch.device('cpu')
-
-
-def atoi(text):
-    return int(text) if text.isdigit() else text
-
-
-def natural_keys(text):
-    '''
-    alist.sort(key=natural_keys) sorts in human order
-    http://nedbatchelder.com/blog/200712/human_sorting.html
-    (See Toothy's implementation in the comments)
-    '''
-    return [atoi(c) for c in re.split(r'(\d+)', text)]
-
-
-def get_model_state_file(checkpoint_dir, zero_stage):
-    if not os.path.isdir(checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
-
-    # there should be only one file
-    if zero_stage <= 2:
-        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
-    elif zero_stage == 3:
-        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
-
-    if not os.path.exists(file):
-        raise FileNotFoundError(f"can't find model states file at '{file}'")
-
-    return file
-
-
-def get_checkpoint_files(checkpoint_dir, glob_pattern):
-    # XXX: need to test that this simple glob rule works for multi-node setup too
-    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
-
-    if len(ckpt_files) == 0:
-        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
-
-    return ckpt_files
-
-
-def get_optim_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
-
-
-def get_model_state_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
-
-
-def parse_model_states(files):
-    zero_model_states = []
-    for file in files:
-        state_dict = torch.load(file, map_location=device)
-
-        if BUFFER_NAMES not in state_dict:
-            raise ValueError(f"{file} is not a model state checkpoint")
-        buffer_names = state_dict[BUFFER_NAMES]
-        if debug:
-            print("Found buffers:", buffer_names)
-
-        # recover just the buffers while restoring them to fp32 if they were saved in fp16
-        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
-        param_shapes = state_dict[PARAM_SHAPES]
-
-        # collect parameters that are included in param_shapes
-        param_names = []
-        for s in param_shapes:
-            for name in s.keys():
-                param_names.append(name)
-
-        # update with frozen parameters
-        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
-        if frozen_param_shapes is not None:
-            if debug:
-                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
-            param_names += list(frozen_param_shapes.keys())
-
-        # handle shared params
-        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
-
-        ds_version = state_dict.get(DS_VERSION, None)
-
-        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
-
-        z_model_state = zero_model_state(buffers=buffers,
-                                         param_shapes=param_shapes,
-                                         shared_params=shared_params,
-                                         ds_version=ds_version,
-                                         frozen_param_shapes=frozen_param_shapes,
-                                         frozen_param_fragments=frozen_param_fragments)
-        zero_model_states.append(z_model_state)
-
-    return zero_model_states
-
-
-def parse_optim_states(files, ds_checkpoint_dir):
-
-    total_files = len(files)
-    state_dicts = []
-    for f in files:
-        state_dict = torch.load(f, map_location=device)
-        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
-        # and also handle the case where it was already removed by another helper script
-        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
-        state_dicts.append(state_dict)
-
-    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
-        raise ValueError(f"{files[0]} is not a zero checkpoint")
-    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
-    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
-
-    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
-    # parameters can be different from data parallelism for non-expert parameters. So we can just
-    # use the max of the partition_count to get the dp world_size.
-
-    if type(world_size) is list:
-        world_size = max(world_size)
-
-    if world_size != total_files:
-        raise ValueError(
-            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
-            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
-        )
-
-    # the groups are named differently in each stage
-    if zero_stage <= 2:
-        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
-    elif zero_stage == 3:
-        fp32_groups_key = FP32_FLAT_GROUPS
-    else:
-        raise ValueError(f"unknown zero stage {zero_stage}")
-
-    if zero_stage <= 2:
-        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
-    elif zero_stage == 3:
-        # if there is more than one param group, there will be multiple flattened tensors - one
-        # flattened tensor per group - for simplicity merge them into a single tensor
-        #
-        # XXX: could make the script more memory efficient for when there are multiple groups - it
-        # will require matching the sub-lists of param_shapes for each param group flattened tensor
-
-        fp32_flat_groups = [
-            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
-        ]
-
-    return zero_stage, world_size, fp32_flat_groups
-
-
-def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
-    """
-    Returns fp32 state_dict reconstructed from ds checkpoint
-
-    Args:
-        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
-
-    """
-    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
-
-    optim_files = get_optim_files(ds_checkpoint_dir)
-    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
-    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
-
-    model_files = get_model_state_files(ds_checkpoint_dir)
-
-    zero_model_states = parse_model_states(model_files)
-    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
-
-    if zero_stage <= 2:
-        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-    elif zero_stage == 3:
-        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-
-
-def _zero2_merge_frozen_params(state_dict, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
-
-    if debug:
-        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
-        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        state_dict[name] = frozen_param_fragments[name]
-
-        if debug:
-            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-
-    # Reconstruction protocol:
-    #
-    # XXX: document this
-
-    if debug:
-        for i in range(world_size):
-            for j in range(len(fp32_flat_groups[0])):
-                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
-
-    # XXX: memory usage doubles here (zero2)
-    num_param_groups = len(fp32_flat_groups[0])
-    merged_single_partition_of_fp32_groups = []
-    for i in range(num_param_groups):
-        merged_partitions = [sd[i] for sd in fp32_flat_groups]
-        full_single_fp32_vector = torch.cat(merged_partitions, 0)
-        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
-    avail_numel = sum(
-        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
-
-    if debug:
-        wanted_params = sum([len(shapes) for shapes in param_shapes])
-        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
-        # not asserting if there is a mismatch due to possible padding
-        print(f"Have {avail_numel} numels to process.")
-        print(f"Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    total_numel = 0
-    total_params = 0
-    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
-        offset = 0
-        avail_numel = full_single_fp32_vector.numel()
-        for name, shape in shapes.items():
-
-            unpartitioned_numel = shape.numel()
-            total_numel += unpartitioned_numel
-            total_params += 1
-
-            if debug:
-                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
-            offset += unpartitioned_numel
-
-        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
-        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
-        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
-        # live optimizer object, so we are checking that the numbers are within the right range
-        align_to = 2 * world_size
-
-        def zero2_align(x):
-            return align_to * math.ceil(x / align_to)
-
-        if debug:
-            print(f"original offset={offset}, avail_numel={avail_numel}")
-
-        offset = zero2_align(offset)
-        avail_numel = zero2_align(avail_numel)
-
-        if debug:
-            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
-
-        # Sanity check
-        if offset != avail_numel:
-            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero2_merge_frozen_params(state_dict, zero_model_states)
-
-    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def zero3_partitioned_param_info(unpartitioned_numel, world_size):
-    remainder = unpartitioned_numel % world_size
-    padding_numel = (world_size - remainder) if remainder else 0
-    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
-    return partitioned_numel, padding_numel
-
-
-def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    if debug:
-        for i in range(world_size):
-            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
-            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in zero_model_states[0].frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
-        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-    avail_numel = fp32_flat_groups[0].numel() * world_size
-    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
-    # param, re-consolidating each param, while dealing with padding if any
-
-    # merge list of dicts, preserving order
-    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
-
-    if debug:
-        for i in range(world_size):
-            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
-
-        wanted_params = len(param_shapes)
-        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
-        # not asserting if there is a mismatch due to possible padding
-        avail_numel = fp32_flat_groups[0].numel() * world_size
-        print(f"Trainable params: Have {avail_numel} numels to process.")
-        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    offset = 0
-    total_numel = 0
-    total_params = 0
-    for name, shape in param_shapes.items():
-
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-        total_params += 1
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-        # XXX: memory usage doubles here
-        state_dict[name] = torch.cat(
-            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
-            0).narrow(0, 0, unpartitioned_numel).view(shape)
-        offset += partitioned_numel
-
-    offset *= world_size
-
-    # Sanity check
-    if offset != avail_numel:
-        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
-
-    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
-    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
-    via a model hub.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
-
-    Returns:
-        - pytorch ``state_dict``
-
-    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
-    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
-    the checkpoint.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
-        # do the training and checkpoint saving
-        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
-        model = model.cpu() # move to cpu
-        model.load_state_dict(state_dict)
-        # submit to model hub or save the model to share with others
-
-    In this example the ``model`` will no longer be usable in the deepspeed context of the same
-    application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
-
-    """
-    if tag is None:
-        latest_path = os.path.join(checkpoint_dir, 'latest')
-        if os.path.isfile(latest_path):
-            with open(latest_path, 'r') as fd:
-                tag = fd.read().strip()
-        else:
-            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
-
-    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
-
-    if not os.path.isdir(ds_checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
-
-    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
-
-
-def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
-    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-    """
-
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-    print(f"Saving fp32 state dict to {output_file}")
-    torch.save(state_dict, output_file)
-
-
-def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
-    """
-    1. Put the provided model to cpu
-    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
-    3. Load it into the provided model
-
-    Args:
-        - ``model``: the model object to update
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-
-    Returns:
-        - ``model`: modified model
-
-    Make sure you have plenty of CPU memory available before you call this function. If you don't
-    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
-    conveniently placed for you in the checkpoint folder.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
-        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
-        # submit to model hub or save the model to share with others
-
-    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
-    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    """
-    logger.info(f"Extracting fp32 weights")
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-
-    logger.info(f"Overwriting model with fp32 weights")
-    model = model.cpu()
-    model.load_state_dict(state_dict, strict=False)
-
-    return model
-
-
-if __name__ == "__main__":
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument("checkpoint_dir",
-                        type=str,
-                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
-    parser.add_argument(
-        "output_file",
-        type=str,
-        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
-    parser.add_argument("-t",
-                        "--tag",
-                        type=str,
-                        default=None,
-                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
-    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
-    args = parser.parse_args()
-
-    debug = args.debug
-
-    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
diff --git a/latest b/latest
deleted file mode 100644
index 9b4dc801e3fb152ef5c0ee60d309c705a9b01564..0000000000000000000000000000000000000000
--- a/latest
+++ /dev/null
@@ -1 +0,0 @@
-global_step50
\ No newline at end of file
diff --git a/rng_state_1.pth b/rng_state_1.pth
deleted file mode 100644
index 6b2ef88173fde17f2b3e738a28446f89a0528a96..0000000000000000000000000000000000000000
--- a/rng_state_1.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7e34eb456d2d003a2839f2daa9425e99bdd79ed7e24a1de9fc7d5738476bfb4b
-size 17655
diff --git a/rng_state_2.pth b/rng_state_2.pth
deleted file mode 100644
index 7b118d52a3006aea6c44f23f94c5568d1fb0a2f3..0000000000000000000000000000000000000000
--- a/rng_state_2.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b374af4a2765d8771cee7a72921d3c2e438b9bee34f0b2d098ce6071afeb65e4
-size 17655
diff --git a/rng_state_3.pth b/rng_state_3.pth
deleted file mode 100644
index 3f6fd9aa58eb1d5815ca991134531a3280601900..0000000000000000000000000000000000000000
--- a/rng_state_3.pth
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5df75d8477fcc69c7abb03025313915ebfe3ac18c54a7c57aaa455c0099e13e5
-size 17655
diff --git a/zero_to_fp32.py b/zero_to_fp32.py
deleted file mode 100755
index c98caae31534368be22b67fc4ae906836c992a8d..0000000000000000000000000000000000000000
--- a/zero_to_fp32.py
+++ /dev/null
@@ -1,587 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright (c) Microsoft Corporation.
-# SPDX-License-Identifier: Apache-2.0
-
-# DeepSpeed Team
-
-# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
-# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
-# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
-# application.
-#
-# example: python zero_to_fp32.py . pytorch_model.bin
-
-import argparse
-import torch
-import glob
-import math
-import os
-import re
-from collections import OrderedDict
-from dataclasses import dataclass
-
-# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
-# DeepSpeed data structures it has to be available in the current python environment.
-from deepspeed.utils import logger
-from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
-                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
-                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
-
-
-@dataclass
-class zero_model_state:
-    buffers: dict()
-    param_shapes: dict()
-    shared_params: list
-    ds_version: int
-    frozen_param_shapes: dict()
-    frozen_param_fragments: dict()
-
-
-debug = 0
-
-# load to cpu
-device = torch.device('cpu')
-
-
-def atoi(text):
-    return int(text) if text.isdigit() else text
-
-
-def natural_keys(text):
-    '''
-    alist.sort(key=natural_keys) sorts in human order
-    http://nedbatchelder.com/blog/200712/human_sorting.html
-    (See Toothy's implementation in the comments)
-    '''
-    return [atoi(c) for c in re.split(r'(\d+)', text)]
-
-
-def get_model_state_file(checkpoint_dir, zero_stage):
-    if not os.path.isdir(checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
-
-    # there should be only one file
-    if zero_stage <= 2:
-        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
-    elif zero_stage == 3:
-        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
-
-    if not os.path.exists(file):
-        raise FileNotFoundError(f"can't find model states file at '{file}'")
-
-    return file
-
-
-def get_checkpoint_files(checkpoint_dir, glob_pattern):
-    # XXX: need to test that this simple glob rule works for multi-node setup too
-    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
-
-    if len(ckpt_files) == 0:
-        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
-
-    return ckpt_files
-
-
-def get_optim_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
-
-
-def get_model_state_files(checkpoint_dir):
-    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
-
-
-def parse_model_states(files):
-    zero_model_states = []
-    for file in files:
-        state_dict = torch.load(file, map_location=device)
-
-        if BUFFER_NAMES not in state_dict:
-            raise ValueError(f"{file} is not a model state checkpoint")
-        buffer_names = state_dict[BUFFER_NAMES]
-        if debug:
-            print("Found buffers:", buffer_names)
-
-        # recover just the buffers while restoring them to fp32 if they were saved in fp16
-        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
-        param_shapes = state_dict[PARAM_SHAPES]
-
-        # collect parameters that are included in param_shapes
-        param_names = []
-        for s in param_shapes:
-            for name in s.keys():
-                param_names.append(name)
-
-        # update with frozen parameters
-        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
-        if frozen_param_shapes is not None:
-            if debug:
-                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
-            param_names += list(frozen_param_shapes.keys())
-
-        # handle shared params
-        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
-
-        ds_version = state_dict.get(DS_VERSION, None)
-
-        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
-
-        z_model_state = zero_model_state(buffers=buffers,
-                                         param_shapes=param_shapes,
-                                         shared_params=shared_params,
-                                         ds_version=ds_version,
-                                         frozen_param_shapes=frozen_param_shapes,
-                                         frozen_param_fragments=frozen_param_fragments)
-        zero_model_states.append(z_model_state)
-
-    return zero_model_states
-
-
-def parse_optim_states(files, ds_checkpoint_dir):
-
-    total_files = len(files)
-    state_dicts = []
-    for f in files:
-        state_dict = torch.load(f, map_location=device)
-        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
-        # and also handle the case where it was already removed by another helper script
-        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
-        state_dicts.append(state_dict)
-
-    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
-        raise ValueError(f"{files[0]} is not a zero checkpoint")
-    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
-    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
-
-    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
-    # parameters can be different from data parallelism for non-expert parameters. So we can just
-    # use the max of the partition_count to get the dp world_size.
-
-    if type(world_size) is list:
-        world_size = max(world_size)
-
-    if world_size != total_files:
-        raise ValueError(
-            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
-            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
-        )
-
-    # the groups are named differently in each stage
-    if zero_stage <= 2:
-        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
-    elif zero_stage == 3:
-        fp32_groups_key = FP32_FLAT_GROUPS
-    else:
-        raise ValueError(f"unknown zero stage {zero_stage}")
-
-    if zero_stage <= 2:
-        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
-    elif zero_stage == 3:
-        # if there is more than one param group, there will be multiple flattened tensors - one
-        # flattened tensor per group - for simplicity merge them into a single tensor
-        #
-        # XXX: could make the script more memory efficient for when there are multiple groups - it
-        # will require matching the sub-lists of param_shapes for each param group flattened tensor
-
-        fp32_flat_groups = [
-            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
-        ]
-
-    return zero_stage, world_size, fp32_flat_groups
-
-
-def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
-    """
-    Returns fp32 state_dict reconstructed from ds checkpoint
-
-    Args:
-        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
-
-    """
-    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
-
-    optim_files = get_optim_files(ds_checkpoint_dir)
-    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
-    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
-
-    model_files = get_model_state_files(ds_checkpoint_dir)
-
-    zero_model_states = parse_model_states(model_files)
-    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
-
-    if zero_stage <= 2:
-        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-    elif zero_stage == 3:
-        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
-
-
-def _zero2_merge_frozen_params(state_dict, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
-
-    if debug:
-        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
-        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        state_dict[name] = frozen_param_fragments[name]
-
-        if debug:
-            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-
-    # Reconstruction protocol:
-    #
-    # XXX: document this
-
-    if debug:
-        for i in range(world_size):
-            for j in range(len(fp32_flat_groups[0])):
-                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
-
-    # XXX: memory usage doubles here (zero2)
-    num_param_groups = len(fp32_flat_groups[0])
-    merged_single_partition_of_fp32_groups = []
-    for i in range(num_param_groups):
-        merged_partitions = [sd[i] for sd in fp32_flat_groups]
-        full_single_fp32_vector = torch.cat(merged_partitions, 0)
-        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
-    avail_numel = sum(
-        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
-
-    if debug:
-        wanted_params = sum([len(shapes) for shapes in param_shapes])
-        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
-        # not asserting if there is a mismatch due to possible padding
-        print(f"Have {avail_numel} numels to process.")
-        print(f"Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    total_numel = 0
-    total_params = 0
-    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
-        offset = 0
-        avail_numel = full_single_fp32_vector.numel()
-        for name, shape in shapes.items():
-
-            unpartitioned_numel = shape.numel()
-            total_numel += unpartitioned_numel
-            total_params += 1
-
-            if debug:
-                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
-            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
-            offset += unpartitioned_numel
-
-        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
-        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
-        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
-        # live optimizer object, so we are checking that the numbers are within the right range
-        align_to = 2 * world_size
-
-        def zero2_align(x):
-            return align_to * math.ceil(x / align_to)
-
-        if debug:
-            print(f"original offset={offset}, avail_numel={avail_numel}")
-
-        offset = zero2_align(offset)
-        avail_numel = zero2_align(avail_numel)
-
-        if debug:
-            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
-
-        # Sanity check
-        if offset != avail_numel:
-            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero2_merge_frozen_params(state_dict, zero_model_states)
-
-    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def zero3_partitioned_param_info(unpartitioned_numel, world_size):
-    remainder = unpartitioned_numel % world_size
-    padding_numel = (world_size - remainder) if remainder else 0
-    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
-    return partitioned_numel, padding_numel
-
-
-def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
-    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
-        return
-
-    if debug:
-        for i in range(world_size):
-            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
-            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
-
-        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
-        wanted_params = len(frozen_param_shapes)
-        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
-        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
-        print(f'Frozen params: Have {avail_numel} numels to process.')
-        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
-
-    total_params = 0
-    total_numel = 0
-    for name, shape in zero_model_states[0].frozen_param_shapes.items():
-        total_params += 1
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-
-        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
-        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
-    param_shapes = zero_model_states[0].param_shapes
-    avail_numel = fp32_flat_groups[0].numel() * world_size
-    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
-    # param, re-consolidating each param, while dealing with padding if any
-
-    # merge list of dicts, preserving order
-    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
-
-    if debug:
-        for i in range(world_size):
-            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
-
-        wanted_params = len(param_shapes)
-        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
-        # not asserting if there is a mismatch due to possible padding
-        avail_numel = fp32_flat_groups[0].numel() * world_size
-        print(f"Trainable params: Have {avail_numel} numels to process.")
-        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
-
-    # params
-    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
-    # out-of-core computing solution
-    offset = 0
-    total_numel = 0
-    total_params = 0
-    for name, shape in param_shapes.items():
-
-        unpartitioned_numel = shape.numel()
-        total_numel += unpartitioned_numel
-        total_params += 1
-
-        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
-
-        if debug:
-            print(
-                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
-            )
-
-        # XXX: memory usage doubles here
-        state_dict[name] = torch.cat(
-            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
-            0).narrow(0, 0, unpartitioned_numel).view(shape)
-        offset += partitioned_numel
-
-    offset *= world_size
-
-    # Sanity check
-    if offset != avail_numel:
-        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
-
-    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
-
-
-def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
-    state_dict = OrderedDict()
-
-    # buffers
-    buffers = zero_model_states[0].buffers
-    state_dict.update(buffers)
-    if debug:
-        print(f"added {len(buffers)} buffers")
-
-    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
-
-    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
-
-    # recover shared parameters
-    for pair in zero_model_states[0].shared_params:
-        if pair[1] in state_dict:
-            state_dict[pair[0]] = state_dict[pair[1]]
-
-    return state_dict
-
-
-def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
-    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
-    via a model hub.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
-
-    Returns:
-        - pytorch ``state_dict``
-
-    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
-    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
-    the checkpoint.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
-        # do the training and checkpoint saving
-        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
-        model = model.cpu() # move to cpu
-        model.load_state_dict(state_dict)
-        # submit to model hub or save the model to share with others
-
-    In this example the ``model`` will no longer be usable in the deepspeed context of the same
-    application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
-
-    """
-    if tag is None:
-        latest_path = os.path.join(checkpoint_dir, 'latest')
-        if os.path.isfile(latest_path):
-            with open(latest_path, 'r') as fd:
-                tag = fd.read().strip()
-        else:
-            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
-
-    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
-
-    if not os.path.isdir(ds_checkpoint_dir):
-        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
-
-    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
-
-
-def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
-    """
-    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
-    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
-
-    Args:
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-    """
-
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-    print(f"Saving fp32 state dict to {output_file}")
-    torch.save(state_dict, output_file)
-
-
-def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
-    """
-    1. Put the provided model to cpu
-    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
-    3. Load it into the provided model
-
-    Args:
-        - ``model``: the model object to update
-        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
-        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
-
-    Returns:
-        - ``model`: modified model
-
-    Make sure you have plenty of CPU memory available before you call this function. If you don't
-    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
-    conveniently placed for you in the checkpoint folder.
-
-    A typical usage might be ::
-
-        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
-        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
-        # submit to model hub or save the model to share with others
-
-    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
-    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
-    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
-
-    """
-    logger.info(f"Extracting fp32 weights")
-    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
-
-    logger.info(f"Overwriting model with fp32 weights")
-    model = model.cpu()
-    model.load_state_dict(state_dict, strict=False)
-
-    return model
-
-
-if __name__ == "__main__":
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument("checkpoint_dir",
-                        type=str,
-                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
-    parser.add_argument(
-        "output_file",
-        type=str,
-        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
-    parser.add_argument("-t",
-                        "--tag",
-                        type=str,
-                        default=None,
-                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
-    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
-    args = parser.parse_args()
-
-    debug = args.debug
-
-    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)