diff --git "a/debug.log" "b/debug.log"
--- "a/debug.log"
+++ "b/debug.log"
@@ -1,5 +1,5 @@
-[2026-01-24 12:51:56,817] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:6937] baseline 0.000GB ()
-[2026-01-24 12:51:56,820] [INFO] [axolotl.cli.config.load_cfg:259] [PID:6937] config:
+[2026-01-24 13:25:03,029] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:9359] baseline 0.000GB ()
+[2026-01-24 13:25:03,032] [INFO] [axolotl.cli.config.load_cfg:259] [PID:9359] config:
 {
   "activation_offloading": false,
   "adapter": "lora",
@@ -49,7 +49,7 @@
   ],
   "eval_max_new_tokens": 128,
   "eval_sample_packing": false,
-  "eval_steps": 100,
+  "eval_steps": 50,
   "eval_strategy": "steps",
   "eval_table_size": 0,
   "experimental_skip_move_to_device": true,
@@ -79,10 +79,11 @@
   ],
   "loraplus_lr_embedding": 1e-06,
   "lr_scheduler": "cosine",
+  "max_steps": 650,
   "mean_resizing_embeddings": false,
   "micro_batch_size": 2,
   "model_config_type": "phi3",
-  "num_epochs": 10.0,
+  "num_epochs": 1.0,
   "optimizer": "adamw_torch",
   "otel_metrics_host": "localhost",
   "otel_metrics_port": 8000,
@@ -101,7 +102,7 @@
   "sample_packing_group_size": 100000,
   "save_only_model": false,
   "save_safetensors": true,
-  "save_steps": 200,
+  "save_steps": 100,
   "save_strategy": "steps",
   "sequence_len": 2048,
   "shuffle_before_merging_datasets": false,
@@ -153,119 +154,119 @@
   },
   "wandb_name": "phi4_lora_axolotl",
   "wandb_project": "phi4_african_history",
-  "warmup_steps": 10,
+  "warmup_steps": 20,
   "weight_decay": 0.0,
   "world_size": 1
 }
-[2026-01-24 12:51:58,452] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:6937] EOS: 199999 / <|endoftext|>
-[2026-01-24 12:51:58,452] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:6937] BOS: 199999 / <|endoftext|>
-[2026-01-24 12:51:58,453] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:6937] PAD: 199999 / <|endoftext|>
-[2026-01-24 12:51:58,453] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:6937] UNK: 199999 / <|endoftext|>
-[2026-01-24 12:51:58,453] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:481] [PID:6937] Unable to find prepared dataset in last_run_prepared/89363fb9438bda5d225c172d067e1ebf
-[2026-01-24 12:51:58,453] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:6937] Loading raw datasets...
-[2026-01-24 12:51:58,453] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:6937] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`.
-[2026-01-24 12:52:01,502] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:6937] Loading dataset: DannyAI/African-History-QA-Dataset with base_type: alpaca_chat.load_qa and prompt_style: None
-[2026-01-24 12:52:01,722] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:224] [PID:6937] min_input_len: 52
-[2026-01-24 12:52:01,722] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:226] [PID:6937] max_input_len: 179
+[2026-01-24 13:25:04,559] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:9359] EOS: 199999 / <|endoftext|>
+[2026-01-24 13:25:04,559] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:9359] BOS: 199999 / <|endoftext|>
+[2026-01-24 13:25:04,559] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:9359] PAD: 199999 / <|endoftext|>
+[2026-01-24 13:25:04,559] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:9359] UNK: 199999 / <|endoftext|>
+[2026-01-24 13:25:04,560] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:481] [PID:9359] Unable to find prepared dataset in last_run_prepared/89363fb9438bda5d225c172d067e1ebf
+[2026-01-24 13:25:04,560] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:9359] Loading raw datasets...
+[2026-01-24 13:25:04,560] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:9359] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`.
+[2026-01-24 13:25:06,680] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:9359] Loading dataset: DannyAI/African-History-QA-Dataset with base_type: alpaca_chat.load_qa and prompt_style: None
+[2026-01-24 13:25:06,878] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:224] [PID:9359] min_input_len: 52
+[2026-01-24 13:25:06,879] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:226] [PID:9359] max_input_len: 179
 
Dropping Long Sequences (>2048) (num_proc=9):   0%| | 0/2114 [00:00<?, ? 
Dropping Long Sequences (>2048) (num_proc=9):  11%| | 235/2114 [00:00<00:
Dropping Long Sequences (>2048) (num_proc=9): 100%|█| 2114/2114 [00:00<00
-
Drop Samples with Zero Trainable Tokens (num_proc=9):   0%| | 0/2114 [00:
Drop Samples with Zero Trainable Tokens (num_proc=9):  11%| | 235/2114 [0
Drop Samples with Zero Trainable Tokens (num_proc=9): 100%|█| 2114/2114 [
+
Drop Samples with Zero Trainable Tokens (num_proc=9):   0%| | 0/2114 [00:
Drop Samples with Zero Trainable Tokens (num_proc=9):  11%| | 235/2114 [0
Drop Samples with Zero Trainable Tokens (num_proc=9): 100%|█| 2114/2114 [
Drop Samples with Zero Trainable Tokens (num_proc=9): 100%|█| 2114/2114 [
 
Add position_id column (Sample Packing) (num_proc=9):   0%| | 0/2114 [00:
Add position_id column (Sample Packing) (num_proc=9):  11%| | 235/2114 [0
Add position_id column (Sample Packing) (num_proc=9): 100%|█| 2114/2114 [
-
Saving the dataset (0/8 shards):   0%|   | 0/2114 [00:00<?, ? examples/s]
Saving the dataset (0/8 shards):  13%|▏| 265/2114 [00:00<00:01, 1084.74 e
Saving the dataset (1/8 shards):  13%|▏| 265/2114 [00:00<00:01, 1084.74 e
Saving the dataset (2/8 shards):  25%|▎| 530/2114 [00:00<00:01, 1084.74 e
Saving the dataset (3/8 shards):  38%|▍| 794/2114 [00:00<00:01, 1084.74 e
Saving the dataset (4/8 shards):  50%|▌| 1058/2114 [00:00<00:00, 1084.74 
Saving the dataset (5/8 shards):  63%|▋| 1322/2114 [00:00<00:00, 1084.74 
Saving the dataset (6/8 shards):  75%|▊| 1586/2114 [00:00<00:00, 1084.74 
Saving the dataset (7/8 shards):  88%|▉| 1850/2114 [00:00<00:00, 1084.74 
Saving the dataset (8/8 shards): 100%|█| 2114/2114 [00:00<00:00, 1084.74 
Saving the dataset (8/8 shards): 100%|█| 2114/2114 [00:00<00:00, 5716.91 
-[2026-01-24 12:52:03,950] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:481] [PID:6937] Unable to find prepared dataset in last_run_prepared/1affaed26259409613b775fd6050f3a2
-[2026-01-24 12:52:03,951] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:6937] Loading raw datasets...
-[2026-01-24 12:52:03,951] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:6937] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`.
-[2026-01-24 12:52:05,236] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:6937] Loading dataset: DannyAI/African-History-QA-Dataset with base_type: alpaca_chat.load_qa and prompt_style: None
-[2026-01-24 12:52:05,446] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:224] [PID:6937] min_input_len: 54
-[2026-01-24 12:52:05,446] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:226] [PID:6937] max_input_len: 169
+
Saving the dataset (0/8 shards):   0%|   | 0/2114 [00:00<?, ? examples/s]
Saving the dataset (0/8 shards):  13%|▏| 265/2114 [00:00<00:01, 1232.71 e
Saving the dataset (1/8 shards):  13%|▏| 265/2114 [00:00<00:01, 1232.71 e
Saving the dataset (2/8 shards):  25%|▎| 529/2114 [00:00<00:01, 1232.71 e
Saving the dataset (3/8 shards):  38%|▍| 794/2114 [00:00<00:01, 1232.71 e
Saving the dataset (4/8 shards):  50%|▌| 1058/2114 [00:00<00:00, 1232.71 
Saving the dataset (5/8 shards):  63%|▋| 1322/2114 [00:00<00:00, 1232.71 
Saving the dataset (6/8 shards):  75%|▊| 1586/2114 [00:00<00:00, 1232.71 
Saving the dataset (7/8 shards):  88%|▉| 1850/2114 [00:00<00:00, 1232.71 
Saving the dataset (8/8 shards): 100%|█| 2114/2114 [00:00<00:00, 1232.71 
Saving the dataset (8/8 shards): 100%|█| 2114/2114 [00:00<00:00, 5928.96 
+[2026-01-24 13:25:09,029] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:481] [PID:9359] Unable to find prepared dataset in last_run_prepared/1affaed26259409613b775fd6050f3a2
+[2026-01-24 13:25:09,030] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:9359] Loading raw datasets...
+[2026-01-24 13:25:09,030] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:9359] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`.
+[2026-01-24 13:25:10,146] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:9359] Loading dataset: DannyAI/African-History-QA-Dataset with base_type: alpaca_chat.load_qa and prompt_style: None
+[2026-01-24 13:25:10,349] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:224] [PID:9359] min_input_len: 54
+[2026-01-24 13:25:10,349] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:226] [PID:9359] max_input_len: 169
 
Dropping Long Sequences (>2048) (num_proc=9):   0%| | 0/200 [00:00<?, ? e
Dropping Long Sequences (>2048) (num_proc=9):  12%| | 23/200 [00:00<00:01
Dropping Long Sequences (>2048) (num_proc=9): 100%|█| 200/200 [00:00<00:0
 
Drop Samples with Zero Trainable Tokens (num_proc=9):   0%| | 0/200 [00:0
Drop Samples with Zero Trainable Tokens (num_proc=9):  12%| | 23/200 [00:
Drop Samples with Zero Trainable Tokens (num_proc=9): 100%|█| 200/200 [00
-
Add position_id column (Sample Packing) (num_proc=9):   0%| | 0/200 [00:0
Add position_id column (Sample Packing) (num_proc=9):  11%| | 22/200 [00:
Add position_id column (Sample Packing) (num_proc=9): 100%|█| 200/200 [00
-
Saving the dataset (0/1 shards):   0%|    | 0/200 [00:00<?, ? examples/s]
Saving the dataset (0/1 shards): 100%|█| 200/200 [00:00<00:00, 1601.52 ex
Saving the dataset (1/1 shards): 100%|█| 200/200 [00:00<00:00, 1601.52 ex
Saving the dataset (1/1 shards): 100%|█| 200/200 [00:00<00:00, 928.58 exa
-[2026-01-24 12:52:07,313] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:417] [PID:6937] total_num_tokens: 205_770
-[2026-01-24 12:52:07,348] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:435] [PID:6937] `total_supervised_tokens: 94_469`
-[2026-01-24 12:52:07,377] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:6937] Using single process for pack_parallel, running sequentially.
-[2026-01-24 12:52:08,322] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:6937] Using single process for pack_parallel, running sequentially.
-[2026-01-24 12:52:08,510] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.18906092643737793
-[2026-01-24 12:52:08,511] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:6937] Using single process for pack_parallel, running sequentially.
-[2026-01-24 12:52:08,705] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.19425106048583984
-[2026-01-24 12:52:08,705] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:6937] Using single process for pack_parallel, running sequentially.
-[2026-01-24 12:52:08,914] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.20883750915527344
-[2026-01-24 12:52:08,914] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:6937] Using single process for pack_parallel, running sequentially.
-[2026-01-24 12:52:09,117] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.20302033424377441
-[2026-01-24 12:52:09,156] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:438] [PID:6937] gather_len_batches: [51]
-[2026-01-24 12:52:09,156] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:494] [PID:6937] data_loader_len: 12
-[2026-01-24 12:52:09,156] [INFO] [axolotl.utils.trainer.calc_sample_packing_eff_est:510] [PID:6937] sample_packing_eff_est across ranks: [0.9850356158088235]
-[2026-01-24 12:52:09,156] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:522] [PID:6937] sample_packing_eff_est: 0.99
-[2026-01-24 12:52:09,156] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:533] [PID:6937] total_num_steps: 120
-[2026-01-24 12:52:09,157] [INFO] [axolotl.utils.data.sft._prepare_standard_dataset:121] [PID:6937] Maximum number of steps set at 120
-[2026-01-24 12:52:09,212] [DEBUG] [axolotl.train.setup_model_and_tokenizer:70] [PID:6937] loading tokenizer... microsoft/Phi-4-mini-instruct
-[2026-01-24 12:52:10,586] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:6937] EOS: 199999 / <|endoftext|>
-[2026-01-24 12:52:10,586] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:6937] BOS: 199999 / <|endoftext|>
-[2026-01-24 12:52:10,586] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:6937] PAD: 199999 / <|endoftext|>
-[2026-01-24 12:52:10,586] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:6937] UNK: 199999 / <|endoftext|>
-[2026-01-24 12:52:10,586] [DEBUG] [axolotl.train.setup_model_and_tokenizer:82] [PID:6937] Loading model
-[2026-01-24 12:52:10,782] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:6937] Patched Trainer.evaluation_loop with nanmean loss calculation
-[2026-01-24 12:52:10,788] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:6937] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation
-[2026-01-24 12:52:10,789] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:345] [PID:6937] Applying multipack dataloader patch for sample packing...
-
Loading checkpoint shards:   0%|                   | 0/2 [00:00<?, ?it/s]
Loading checkpoint shards:  50%|█████▌     | 1/2 [00:00<00:00,  1.24it/s]
Loading checkpoint shards: 100%|███████████| 2/2 [00:01<00:00,  1.55it/s]
Loading checkpoint shards: 100%|███████████| 2/2 [00:01<00:00,  1.49it/s]
-[2026-01-24 12:52:16,277] [INFO] [axolotl.loaders.model._configure_embedding_dtypes:347] [PID:6937] Converting modules to torch.bfloat16
-[2026-01-24 12:52:17,178] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:6937] Memory usage after model load 0.000GB ()
+
Add position_id column (Sample Packing) (num_proc=9):   0%| | 0/200 [00:0
Add position_id column (Sample Packing) (num_proc=9):  12%| | 23/200 [00:
Add position_id column (Sample Packing) (num_proc=9): 100%|█| 200/200 [00
+
Saving the dataset (0/1 shards):   0%|    | 0/200 [00:00<?, ? examples/s]
Saving the dataset (0/1 shards): 100%|█| 200/200 [00:00<00:00, 1565.14 ex
Saving the dataset (1/1 shards): 100%|█| 200/200 [00:00<00:00, 1565.14 ex
Saving the dataset (1/1 shards): 100%|█| 200/200 [00:00<00:00, 905.63 exa
+[2026-01-24 13:25:12,212] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:417] [PID:9359] total_num_tokens: 205_770
+[2026-01-24 13:25:12,246] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:435] [PID:9359] `total_supervised_tokens: 94_469`
+[2026-01-24 13:25:12,274] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:9359] Using single process for pack_parallel, running sequentially.
+[2026-01-24 13:25:13,265] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:9359] Using single process for pack_parallel, running sequentially.
+[2026-01-24 13:25:13,472] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.20749449729919434
+[2026-01-24 13:25:13,473] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:9359] Using single process for pack_parallel, running sequentially.
+[2026-01-24 13:25:13,678] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.20551633834838867
+[2026-01-24 13:25:13,679] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:9359] Using single process for pack_parallel, running sequentially.
+[2026-01-24 13:25:13,870] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.19202852249145508
+[2026-01-24 13:25:13,871] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:9359] Using single process for pack_parallel, running sequentially.
+[2026-01-24 13:25:14,081] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.21112775802612305
+[2026-01-24 13:25:14,120] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:438] [PID:9359] gather_len_batches: [51]
+[2026-01-24 13:25:14,120] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:494] [PID:9359] data_loader_len: 12
+[2026-01-24 13:25:14,120] [INFO] [axolotl.utils.trainer.calc_sample_packing_eff_est:510] [PID:9359] sample_packing_eff_est across ranks: [0.9850356158088235]
+[2026-01-24 13:25:14,120] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:522] [PID:9359] sample_packing_eff_est: 0.99
+[2026-01-24 13:25:14,120] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:533] [PID:9359] total_num_steps: 12
+[2026-01-24 13:25:14,121] [INFO] [axolotl.utils.data.sft._prepare_standard_dataset:121] [PID:9359] Maximum number of steps set at 12
+[2026-01-24 13:25:14,169] [DEBUG] [axolotl.train.setup_model_and_tokenizer:70] [PID:9359] loading tokenizer... microsoft/Phi-4-mini-instruct
+[2026-01-24 13:25:15,574] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:9359] EOS: 199999 / <|endoftext|>
+[2026-01-24 13:25:15,574] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:9359] BOS: 199999 / <|endoftext|>
+[2026-01-24 13:25:15,574] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:9359] PAD: 199999 / <|endoftext|>
+[2026-01-24 13:25:15,574] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:9359] UNK: 199999 / <|endoftext|>
+[2026-01-24 13:25:15,574] [DEBUG] [axolotl.train.setup_model_and_tokenizer:82] [PID:9359] Loading model
+[2026-01-24 13:25:15,776] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:9359] Patched Trainer.evaluation_loop with nanmean loss calculation
+[2026-01-24 13:25:15,781] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:9359] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation
+[2026-01-24 13:25:15,782] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:345] [PID:9359] Applying multipack dataloader patch for sample packing...
+
Loading checkpoint shards:   0%|                   | 0/2 [00:00<?, ?it/s]
Loading checkpoint shards: 100%|███████████| 2/2 [00:00<00:00, 66.49it/s]
+[2026-01-24 13:25:17,426] [INFO] [axolotl.loaders.model._configure_embedding_dtypes:347] [PID:9359] Converting modules to torch.bfloat16
+[2026-01-24 13:25:18,248] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:9359] Memory usage after model load 0.000GB ()
 trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
-[2026-01-24 12:52:17,232] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:6937] after adapters 0.000GB ()
-[2026-01-24 12:52:47,260] [INFO] [axolotl.train.save_initial_configs:413] [PID:6937] Pre-saving adapter config to ./phi4_african_history_lora_out...
-[2026-01-24 12:52:47,261] [INFO] [axolotl.train.save_initial_configs:417] [PID:6937] Pre-saving tokenizer to ./phi4_african_history_lora_out...
-[2026-01-24 12:52:47,558] [INFO] [axolotl.train.save_initial_configs:422] [PID:6937] Pre-saving model config to ./phi4_african_history_lora_out...
-[2026-01-24 12:52:47,563] [INFO] [axolotl.train.execute_training:212] [PID:6937] Starting trainer...
-[2026-01-24 12:52:49,594] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.7298610210418701
-[2026-01-24 12:52:50,250] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.6557362079620361
-[2026-01-24 12:52:50,759] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.5084567070007324
-[2026-01-24 12:52:51,268] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:6937] generate_batches time: 0.5085635185241699
-[2026-01-24 12:52:51,268] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:438] [PID:6937] gather_len_batches: [51]
+[2026-01-24 13:25:18,300] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:9359] after adapters 0.000GB ()
+[2026-01-24 13:25:26,968] [INFO] [axolotl.train.save_initial_configs:413] [PID:9359] Pre-saving adapter config to ./phi4_african_history_lora_out...
+[2026-01-24 13:25:26,968] [INFO] [axolotl.train.save_initial_configs:417] [PID:9359] Pre-saving tokenizer to ./phi4_african_history_lora_out...
+[2026-01-24 13:25:27,149] [INFO] [axolotl.train.save_initial_configs:422] [PID:9359] Pre-saving model config to ./phi4_african_history_lora_out...
+[2026-01-24 13:25:27,153] [INFO] [axolotl.train.execute_training:212] [PID:9359] Starting trainer...
+[2026-01-24 13:25:28,505] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.47576427459716797
+[2026-01-24 13:25:28,983] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.477489709854126
+[2026-01-24 13:25:29,462] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.47864699363708496
+[2026-01-24 13:25:29,967] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:9359] generate_batches time: 0.5052089691162109
+[2026-01-24 13:25:29,968] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:438] [PID:9359] gather_len_batches: [51]
 wandb: [wandb.login()] Loaded credentials for https://api.wandb.ai from /root/.netrc.
 wandb: Currently logged in as: dannyai (dannyai-danny-the-analyst) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
 wandb: ⢿ Waiting for wandb.init()...
-
m
wandb: ⣻ setting up run ehuutbte (0.1s)
-
m
wandb: ⣽ setting up run ehuutbte (0.1s)
+
m
wandb: ⣻ setting up run 80snmd5o (0.2s)
+
m
wandb: ⣽ setting up run 80snmd5o (0.2s)
 
m
wandb: Tracking run with wandb version 0.24.0
-wandb: Run data is saved locally in /workspace/wandb/run-20260124_125251-ehuutbte
+wandb: Run data is saved locally in /workspace/wandb/run-20260124_132530-80snmd5o
 wandb: Run `wandb offline` to turn off syncing.
 wandb: Syncing run phi4_lora_axolotl
 wandb: ⭐️ View project at https://wandb.ai/dannyai-danny-the-analyst/phi4_african_history
-wandb: 🚀 View run at https://wandb.ai/dannyai-danny-the-analyst/phi4_african_history/runs/ehuutbte
+wandb: 🚀 View run at https://wandb.ai/dannyai-danny-the-analyst/phi4_african_history/runs/80snmd5o
 wandb: Detected [huggingface_hub.inference] in use.
 wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
 wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
 wandb: WARNING Saving files without folders. If you want to preserve subdirectories pass base_path to wandb.save, i.e. wandb.save("/mnt/folder/file.h5", base_path="/mnt")
 wandb: WARNING Symlinked 1 file into the W&B run directory; call wandb.save again to sync new files.
-[2026-01-24 12:52:54,066] [INFO] [axolotl.utils.callbacks.on_train_begin:757] [PID:6937] The Axolotl config has been saved to the WandB run under files.
-
  0%|                                            | 0/120 [00:00<?, ?it/s][2026-01-24 12:52:54,083] [INFO] [axolotl.core.trainers.base.evaluate:400] [PID:6937] Running evaluation step...
+[2026-01-24 13:25:33,022] [INFO] [axolotl.utils.callbacks.on_train_begin:757] [PID:9359] The Axolotl config has been saved to the WandB run under files.
+
  0%|                                            | 0/650 [00:00<?, ?it/s][2026-01-24 13:25:33,028] [INFO] [axolotl.core.trainers.base.evaluate:400] [PID:9359] Running evaluation step...
 
 
  0%|                                            | 0/100 [00:00<?, ?it/s]
-
  2%|▋                                   | 2/100 [00:00<00:08, 11.63it/s]
+
  2%|▋                                   | 2/100 [00:00<00:08, 11.46it/s]
 
  4%|█▍                                  | 4/100 [00:01<00:32,  2.97it/s]
-
  5%|█▊                                  | 5/100 [00:01<00:37,  2.50it/s]
-
  6%|██▏                                 | 6/100 [00:02<00:40,  2.32it/s]
-
  7%|██▌                                 | 7/100 [00:02<00:42,  2.20it/s]
+
  5%|█▊                                  | 5/100 [00:01<00:38,  2.49it/s]
+
  6%|██▏                                 | 6/100 [00:02<00:40,  2.31it/s]
+
  7%|██▌                                 | 7/100 [00:02<00:42,  2.19it/s]
 
  8%|██▉                                 | 8/100 [00:03<00:43,  2.12it/s]
-
  9%|███▏                                | 9/100 [00:03<00:45,  2.01it/s]
-
 10%|███▌                               | 10/100 [00:04<00:44,  2.00it/s]
-
 11%|███▊                               | 11/100 [00:04<00:44,  1.99it/s]
-
 12%|████▏                              | 12/100 [00:05<00:44,  1.98it/s]
-
 13%|████▌                              | 13/100 [00:05<00:45,  1.92it/s]
-
 14%|████▉                              | 14/100 [00:06<00:44,  1.94it/s]
-
 15%|█████▎                             | 15/100 [00:06<00:43,  1.94it/s]
-
 16%|█████▌                             | 16/100 [00:07<00:43,  1.95it/s]
+
  9%|███▏                                | 9/100 [00:03<00:45,  2.02it/s]
+
 10%|███▌                               | 10/100 [00:04<00:44,  2.01it/s]
+
 11%|███▊                               | 11/100 [00:04<00:44,  2.00it/s]
+
 12%|████▏                              | 12/100 [00:05<00:44,  1.99it/s]
+
 13%|████▌                              | 13/100 [00:05<00:44,  1.94it/s]
+
 14%|████▉                              | 14/100 [00:06<00:44,  1.95it/s]
+
 15%|█████▎                             | 15/100 [00:06<00:43,  1.95it/s]
+
 16%|█████▌                             | 16/100 [00:07<00:42,  1.96it/s]
 
 17%|█████▉                             | 17/100 [00:07<00:43,  1.91it/s]
 
 18%|██████▎                            | 18/100 [00:08<00:42,  1.93it/s]
-
 19%|██████▋                            | 19/100 [00:09<00:41,  1.94it/s]
+
 19%|██████▋                            | 19/100 [00:08<00:41,  1.94it/s]
 
 20%|███████                            | 20/100 [00:09<00:41,  1.95it/s]
 
 21%|███████▎                           | 21/100 [00:10<00:41,  1.91it/s]
 
 22%|███████▋                           | 22/100 [00:10<00:40,  1.93it/s]
 
 23%|████████                           | 23/100 [00:11<00:39,  1.94it/s]
-
 24%|████████▍                          | 24/100 [00:11<00:39,  1.95it/s]
+
 24%|████████▍                          | 24/100 [00:11<00:39,  1.94it/s]
 
 25%|████████▊                          | 25/100 [00:12<00:39,  1.90it/s]
 
 26%|█████████                          | 26/100 [00:12<00:38,  1.93it/s]
-
 27%|█████████▍                         | 27/100 [00:13<00:37,  1.94it/s]
+
 27%|█████████▍                         | 27/100 [00:13<00:37,  1.93it/s]
 
 28%|█████████▊                         | 28/100 [00:13<00:37,  1.94it/s]
 
 29%|██████████▏                        | 29/100 [00:14<00:37,  1.90it/s]
 
 30%|██████████▌                        | 30/100 [00:14<00:36,  1.92it/s]
@@ -287,17 +288,17 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 46%|████████████████                   | 46/100 [00:23<00:28,  1.92it/s]
 
 47%|████████████████▍                  | 47/100 [00:23<00:27,  1.93it/s]
 
 48%|████████████████▊                  | 48/100 [00:24<00:26,  1.94it/s]
-
 49%|█████████████████▏                 | 49/100 [00:24<00:26,  1.90it/s]
+
 49%|█████████████████▏                 | 49/100 [00:24<00:26,  1.89it/s]
 
 50%|█████████████████▌                 | 50/100 [00:25<00:26,  1.92it/s]
 
 51%|█████████████████▊                 | 51/100 [00:25<00:25,  1.93it/s]
-
 52%|██████████████████▏                | 52/100 [00:26<00:24,  1.94it/s]
-
 53%|██████████████████▌                | 53/100 [00:26<00:24,  1.90it/s]
-
 54%|██████████████████▉                | 54/100 [00:27<00:23,  1.92it/s]
+
 52%|██████████████████▏                | 52/100 [00:26<00:24,  1.93it/s]
+
 53%|██████████████████▌                | 53/100 [00:26<00:24,  1.89it/s]
+
 54%|██████████████████▉                | 54/100 [00:27<00:24,  1.92it/s]
 
 55%|███████████████████▎               | 55/100 [00:27<00:23,  1.93it/s]
 
 56%|███████████████████▌               | 56/100 [00:28<00:22,  1.93it/s]
-
 57%|███████████████████▉               | 57/100 [00:28<00:22,  1.90it/s]
+
 57%|███████████████████▉               | 57/100 [00:28<00:22,  1.89it/s]
 
 58%|████████████████████▎              | 58/100 [00:29<00:21,  1.92it/s]
-
 59%|████████████████████▋              | 59/100 [00:29<00:21,  1.92it/s]
+
 59%|████████████████████▋              | 59/100 [00:29<00:21,  1.93it/s]
 
 60%|█████████████████████              | 60/100 [00:30<00:20,  1.93it/s]
 
 61%|█████████████████████▎             | 61/100 [00:30<00:20,  1.89it/s]
 
 62%|█████████████████████▋             | 62/100 [00:31<00:19,  1.91it/s]
@@ -305,28 +306,28 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 64%|██████████████████████▍            | 64/100 [00:32<00:18,  1.93it/s]
 
 65%|██████████████████████▊            | 65/100 [00:32<00:18,  1.89it/s]
 
 66%|███████████████████████            | 66/100 [00:33<00:17,  1.91it/s]
-
 67%|███████████████████████▍           | 67/100 [00:34<00:17,  1.92it/s]
+
 67%|███████████████████████▍           | 67/100 [00:33<00:17,  1.92it/s]
 
 68%|███████████████████████▊           | 68/100 [00:34<00:16,  1.93it/s]
 
 69%|████████████████████████▏          | 69/100 [00:35<00:16,  1.89it/s]
 
 70%|████████████████████████▌          | 70/100 [00:35<00:15,  1.91it/s]
 
 71%|████████████████████████▊          | 71/100 [00:36<00:15,  1.92it/s]
-
 72%|█████████████████████████▏         | 72/100 [00:36<00:14,  1.93it/s]
+
 72%|█████████████████████████▏         | 72/100 [00:36<00:14,  1.92it/s]
 
 73%|█████████████████████████▌         | 73/100 [00:37<00:14,  1.89it/s]
 
 74%|█████████████████████████▉         | 74/100 [00:37<00:13,  1.91it/s]
 
 75%|██████████████████████████▎        | 75/100 [00:38<00:13,  1.92it/s]
-
 76%|██████████████████████████▌        | 76/100 [00:38<00:12,  1.93it/s]
-
 77%|██████████████████████████▉        | 77/100 [00:39<00:12,  1.89it/s]
+
 76%|██████████████████████████▌        | 76/100 [00:38<00:12,  1.92it/s]
+
 77%|██████████████████████████▉        | 77/100 [00:39<00:12,  1.88it/s]
 
 78%|███████████████████████████▎       | 78/100 [00:39<00:11,  1.91it/s]
 
 79%|███████████████████████████▋       | 79/100 [00:40<00:10,  1.92it/s]
 
 80%|████████████████████████████       | 80/100 [00:40<00:10,  1.92it/s]
 
 81%|████████████████████████████▎      | 81/100 [00:41<00:10,  1.88it/s]
-
 82%|████████████████████████████▋      | 82/100 [00:41<00:09,  1.91it/s]
-
 83%|█████████████████████████████      | 83/100 [00:42<00:08,  1.92it/s]
+
 82%|████████████████████████████▋      | 82/100 [00:41<00:09,  1.90it/s]
+
 83%|█████████████████████████████      | 83/100 [00:42<00:08,  1.91it/s]
 
 84%|█████████████████████████████▍     | 84/100 [00:42<00:08,  1.92it/s]
 
 85%|█████████████████████████████▊     | 85/100 [00:43<00:07,  1.88it/s]
 
 86%|██████████████████████████████     | 86/100 [00:43<00:07,  1.90it/s]
 
 87%|██████████████████████████████▍    | 87/100 [00:44<00:06,  1.91it/s]
-
 88%|██████████████████████████████▊    | 88/100 [00:44<00:06,  1.92it/s]
+
 88%|██████████████████████████████▊    | 88/100 [00:45<00:06,  1.92it/s]
 
 89%|███████████████████████████████▏   | 89/100 [00:45<00:05,  1.88it/s]
 
 90%|███████████████████████████████▌   | 90/100 [00:46<00:05,  1.90it/s]
 
 91%|███████████████████████████████▊   | 91/100 [00:46<00:04,  1.91it/s]
@@ -339,51 +340,156 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 98%|██████████████████████████████████▎| 98/100 [00:50<00:01,  1.90it/s]
 
 99%|██████████████████████████████████▋| 99/100 [00:50<00:00,  1.91it/s]
 
100%|██████████████████████████████████| 100/100 [00:51<00:00,  1.85it/s]
                                                                         
-
                                                                         
{'eval_loss': 2.133481979370117, 'eval_runtime': 53.351, 'eval_samples_per_second': 3.749, 'eval_steps_per_second': 1.874, 'eval_ppl': 8.44422, 'memory/max_active (GiB)': 14.82, 'memory/max_allocated (GiB)': 14.82, 'memory/device_reserved (GiB)': 15.37, 'epoch': 0}
-
  0%|                                            | 0/120 [00:53<?, ?it/s]
+
                                                                         
{'eval_loss': 2.118363380432129, 'eval_runtime': 55.7531, 'eval_samples_per_second': 3.587, 'eval_steps_per_second': 1.794, 'eval_ppl': 8.31751, 'memory/max_active (GiB)': 14.82, 'memory/max_allocated (GiB)': 14.82, 'memory/device_reserved (GiB)': 15.37, 'epoch': 0}
+
  0%|                                            | 0/650 [00:55<?, ?it/s]
 
100%|██████████████████████████████████| 100/100 [00:51<00:00,  1.85it/s]
-
                                                                         
  1%|▎                                 | 1/120 [01:03<2:06:12, 63.63s/it]
  2%|▌                                 | 2/120 [01:11<1:01:02, 31.04s/it]
  2%|▉                                   | 3/120 [01:20<40:13, 20.63s/it]
  3%|█▏                                  | 4/120 [01:28<30:26, 15.74s/it]
  4%|█▌                                  | 5/120 [01:36<24:59, 13.04s/it]
                                                                         
{'loss': 6.0041, 'grad_norm': 0.5430848002433777, 'learning_rate': 8.000000000000001e-06, 'ppl': 405.08625, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 233.90234375, 'tokens/total': 81920, 'tokens/trainable': 36719, 'epoch': 0.39}
-
  4%|█▌                                  | 5/120 [01:36<24:59, 13.04s/it]
  5%|█▊                                  | 6/120 [01:44<21:41, 11.41s/it]
  6%|██                                  | 7/120 [01:53<19:33, 10.38s/it]
  7%|██▍                                 | 8/120 [02:01<18:06,  9.70s/it]
  8%|██▋                                 | 9/120 [02:09<17:06,  9.25s/it]
  8%| | 10/120 [02:17<16:23,  8                                          
                               
{'loss': 6.0771, 'grad_norm': 0.6335939168930054, 'learning_rate': 1.8e-05, 'ppl': 435.76365, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 224.22862243652344, 'tokens/total': 163840, 'tokens/trainable': 74404, 'epoch': 0.78}
-
  8%| | 10/120 [02:17<16:23,  8
  9%|█████▌                                                       | 11/120 [02:26<15:51,  8.73s/it]
 10%|████████████▍                                                                                                               | 12/120 [02:34<15:26,  8.58s/it]
 11%|█████████████▍                                                                                                              | 13/120 [02:40<14:02,  7.87s/it]
 12%|████                               | 14/120 [02:50<15:11,  8.60s/it]                                                                                         
 12%|███████▋                                                     | 15/120 [02:59<14:52,  8.50s/it]
                                                                                                   
{'loss': 6.0242, 'grad_norm': 0.7408789992332458, 'learning_rate': 1.99348173534855e-05, 'ppl': 413.31086, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 234.21719360351562, 'tokens/total': 241664, 'tokens/trainable': 109470, 'epoch': 1.16}
-
 12%|███████▋                                                     | 15/120 [02:59<14:52,  8.50s/it]
 13%|████████▏                                                    | 16/120 [03:07<14:37,  8.44s/it]
 14%|████████▋                                                    | 17/120 [03:15<14:24,  8.39s/it]
 15%|█████████▏                                                   | 18/120 [03:23<14:12,  8.35s/it]
 16%|█████████▋                                                   | 19/120 [03:32<14:01,  8.34s/it]
 17%|██████████▏                                                  | 20/120 [03:40<13:51,  8.32s/it]
                                                                                                   
{'loss': 6.0268, 'grad_norm': 0.872515857219696, 'learning_rate': 1.9671468547019575e-05, 'ppl': 414.38687, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 235.6763916015625, 'tokens/total': 323584, 'tokens/trainable': 146755, 'epoch': 1.55}
-
 17%|██████████▏                                                  | 20/120 [03:40<13:51,  8.32s/it]
 18%|██████████▋                                                  | 21/120 [03:48<13:42,  8.30s/it]
 18%|██████▍                            | 22/120 [03:57<13:33,  8.30s/it]                          
 19%|██████▋                            | 23/120 [04:05<13:24,  8.29s/it]
 20%|███████                            | 24/120 [04:13<13:15,  8.29s/it]
 21%|███████▎                           | 25/120 [04:21<13:07,  8.29s/it]
                                                                         
{'loss': 5.8922, 'grad_norm': 1.1277377605438232, 'learning_rate': 1.92112365311485e-05, 'ppl': 362.20125, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 232.94598388671875, 'tokens/total': 405504, 'tokens/trainable': 184142, 'epoch': 1.94}
-
 21%|███████▎                           | 25/120 [04:21<13:07,  8.29s/it]
 22%|███████▌                           | 26/120 [04:28<12:02,  7.69s/it]
 22%|███████▉                           | 27/120 [04:39<13:27,  8.68s/it]
 23%|████████▏                          | 28/120 [04:47<13:07,  8.56s/it]
 24%|████████▍                          | 29/120 [04:54<12:26,  8.20s/it]
 25%|████████▊                          | 30/120 [05:03<12:19,  8.22s/it]
                                                                         
{'loss': 5.8454, 'grad_norm': 1.1392470598220825, 'learning_rate': 1.856349030251589e-05, 'ppl': 345.64077, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 229.9932861328125, 'tokens/total': 483328, 'tokens/trainable': 218643, 'epoch': 2.31}
-
 25%|████████▊                          | 30/120 [05:03<12:19,  8.22s/it]
 26%|█████████                          | 31/120 [05:11<12:12,  8.23s/it]
 27%|█████████▎                         | 32/120 [05:19<12:05,  8.24s/it]
 28%|█████████▋                         | 33/120 [05:27<11:57,  8.25s/it]
 28%|█████████▉                         | 34/120 [05:36<11:50,  8.26s/it]
 29%|██████████▏                        | 35/120 [05:44<11:41,  8.26s/it]
                                                                         
{'loss': 5.7195, 'grad_norm': 1.2083805799484253, 'learning_rate': 1.7741416106390828e-05, 'ppl': 304.75251, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 218.5542449951172, 'tokens/total': 565248, 'tokens/trainable': 255915, 'epoch': 2.71}
-
 29%|██████████▏                        | 35/120 [05:44<11:41,  8.26s/it]
 30%|██████████▌                        | 36/120 [05:52<11:33,  8.26s/it]
 31%|██████████▊                        | 37/120 [06:01<11:25,  8.26s/it]
 32%|███████████                        | 38/120 [06:09<11:17,  8.26s/it]
 32%|███████████▍                       | 39/120 [06:15<10:20,  7.66s/it]
 33%|███████████▋                       | 40/120 [06:25<11:15,  8.44s/it]
                                                                         
{'loss': 5.6459, 'grad_norm': 1.3146638870239258, 'learning_rate': 1.6761749002740195e-05, 'ppl': 283.12826, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 204.4563751220703, 'tokens/total': 643072, 'tokens/trainable': 290592, 'epoch': 3.08}
-
 33%|███████████▋                       | 40/120 [06:25<11:15,  8.44s/it]
 34%|███████████▉                       | 41/120 [06:34<11:03,  8.39s/it]
 35%|█████████���██▎                      | 42/120 [06:42<10:52,  8.36s/it]
 36%|████████████▌                      | 43/120 [06:50<10:42,  8.34s/it]
 37%|████████████▊                      | 44/120 [06:58<10:32,  8.32s/it]
 38%|█████████████▏                     | 45/120 [07:07<10:22,  8.31s/it]
                                                                         
{'loss': 5.5502, 'grad_norm': 1.3570324182510376, 'learning_rate': 1.5644432188667695e-05, 'ppl': 257.28901, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 225.54989624023438, 'tokens/total': 724992, 'tokens/trainable': 327768, 'epoch': 3.47}
-
 38%|█████████████▏                     | 45/120 [07:07<10:22,  8.31s/it]
 38%|█████████████▍                     | 46/120 [07:15<10:13,  8.30s/it]
 39%|█████████████▋                     | 47/120 [07:23<10:04,  8.29s/it]
 40%|██████████████                     | 48/120 [07:32<09:56,  8.28s/it]
 41%|██████████████▎                    | 49/120 [07:39<09:27,  8.00s/it]
 42%|██████████████▌                    | 50/120 [07:47<09:25,  8.07s/it]
                                                                         
{'loss': 5.4407, 'grad_norm': 1.4176502227783203, 'learning_rate': 1.4412211012432213e-05, 'ppl': 230.60355, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 225.1769561767578, 'tokens/total': 806912, 'tokens/trainable': 365516, 'epoch': 3.86}
-
 42%|██████████████▌                    | 50/120 [07:47<09:25,  8.07s/it]
 42%|██████████████▉                    | 51/120 [07:55<09:20,  8.13s/it]
 43%|███████████████▏                   | 52/120 [08:02<08:34,  7.57s/it]
 44%|███████████████▍                   | 53/120 [08:12<09:17,  8.33s/it]
 45%|███████████████▊                   | 54/120 [08:20<09:08,  8.31s/it]
 46%|████████████████                   | 55/120 [08:28<08:59,  8.30s/it]
                                                                         
{'loss': 5.3679, 'grad_norm': 1.3387575149536133, 'learning_rate': 1.3090169943749475e-05, 'ppl': 214.41213, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 219.9198455810547, 'tokens/total': 884736, 'tokens/trainable': 400250, 'epoch': 4.24}
-
 46%|████████████████                   | 55/120 [08:28<08:59,  8.30s/it]
 47%|████████████████▎                  | 56/120 [08:37<08:50,  8.29s/it]
 48%|████████████████▋                  | 57/120 [08:45<08:41,  8.28s/it]
 48%|████████████████▉                  | 58/120 [08:53<08:33,  8.28s/it]
 49%|█████████████████▏                 | 59/120 [09:01<08:24,  8.27s/it]
 50%|█████████████████▌                 | 60/120 [09:10<08:16,  8.27s/it]
                                                                         
{'loss': 5.3246, 'grad_norm': 1.3205095529556274, 'learning_rate': 1.170522192632624e-05, 'ppl': 205.32621, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 249.20303344726562, 'tokens/total': 966656, 'tokens/trainable': 437702, 'epoch': 4.63}
-
 50%|█████████████████▌                 | 60/120 [09:10<08:16,  8.27s/it]
 51%|█████████████████▊                 | 61/120 [09:18<08:08,  8.27s/it]
 52%|██████████████████                 | 62/120 [09:26<07:59,  8.27s/it]
 52%|██████████████████▍                | 63/120 [09:34<07:51,  8.27s/it]
 53%|██████████████████▋                | 64/120 [09:43<07:42,  8.27s/it]
 54%|██████████████████▉                | 65/120 [09:49<07:01,  7.66s/it]
                                                                         
{'loss': 5.1853, 'grad_norm': 1.1749951839447021, 'learning_rate': 1.0285560507936962e-05, 'ppl': 178.62703, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 215.25717163085938, 'tokens/total': 1044480, 'tokens/trainable': 472345, 'epoch': 5.0}
-
 54%|██████████████████▉                | 65/120 [09:49<07:01,  7.66s/it]
 55%|███████████████████▎               | 66/120 [09:59<07:34,  8.42s/it]
 56%|███████████████████▌               | 67/120 [10:07<07:23,  8.38s/it]
 57%|███████████████████▊               | 68/120 [10:16<07:13,  8.34s/it]
 57%|████████████████████▏              | 69/120 [10:24<07:04,  8.32s/it]
 58%|████████████████████▍              | 70/120 [10:32<06:55,  8.30s/it]
                                                                         
{'loss': 5.1466, 'grad_norm': 1.2770065069198608, 'learning_rate': 8.860085901094595e-06, 'ppl': 171.84622, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 232.2105255126953, 'tokens/total': 1126400, 'tokens/trainable': 509718, 'epoch': 5.39}
-
 58%|████████████████████▍              | 70/120 [10:32<06:55,  8.30s/it]
 59%|████████████████████▋              | 71/120 [10:40<06:46,  8.30s/it]
 60%|█████████████████████              | 72/120 [10:49<06:37,  8.29s/it]
 61%|█████████████████████▎             | 73/120 [10:57<06:29,  8.28s/it]
 62%|█████████████████████▌             | 74/120 [11:04<06:07,  8.00s/it]
 62%|█████████████████████▉             | 75/120 [11:13<06:03,  8.08s/it]
                                                                         
{'loss': 5.063, 'grad_norm': 1.2473822832107544, 'learning_rate': 7.4578166580651335e-06, 'ppl': 158.064, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 233.98727416992188, 'tokens/total': 1208320, 'tokens/trainable': 547102, 'epoch': 5.78}
-
 62%|█████████████████████▉             | 75/120 [11:13<06:03,  8.08s/it]
 63%|██████████████████████▏            | 76/120 [11:21<05:58,  8.14s/it]
 64%|██████████████████████▍            | 77/120 [11:29<05:51,  8.18s/it]
 65%|██████████████████████▊            | 78/120 [11:35<05:19,  7.60s/it]
 66%|███████████████████████            | 79/120 [11:46<05:44,  8.40s/it]
 67%|███████████████████████▎           | 80/120 [11:54<05:34,  8.36s/it]
                                                                         
{'loss': 5.0777, 'grad_norm': 1.351817011833191, 'learning_rate': 6.107298936826086e-06, 'ppl': 160.4047, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 222.9584503173828, 'tokens/total': 1286144, 'tokens/trainable': 581733, 'epoch': 6.16}
-
 67%|███████████████████████▎           | 80/120 [11:54<05:34,  8.36s/it]
 68%|███████████████████████▋           | 81/120 [12:02<05:24,  8.33s/it]
 68%|███████████████████████▉           | 82/120 [12:10<05:15,  8.31s/it]
 69%|████████████████████████▏          | 83/120 [12:19<05:07,  8.30s/it]
 70%|████████████████████████▌          | 84/120 [12:27<04:58,  8.29s/it]
 71%|████████████████████████▊          | 85/120 [12:35<04:49,  8.28s/it]
                                                                         
{'loss': 5.0529, 'grad_norm': 1.2326539754867554, 'learning_rate': 4.836025383610382e-06, 'ppl': 156.47559, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 211.7113494873047, 'tokens/total': 1368064, 'tokens/trainable': 618918, 'epoch': 6.55}
-
 71%|████████████████████████▊          | 85/120 [12:35<04:49,  8.28s/it]
 72%|█████████████████████████          | 86/120 [12:43<04:41,  8.27s/it]
 72%|█████████████████████████▍         | 87/120 [12:52<04:32,  8.27s/it]
 73%|█████████████████████████▋         | 88/120 [12:59<04:15,  7.99s/it]
 74%|█████████████████████████▉         | 89/120 [13:07<04:10,  8.07s/it]
 75%|██████████████████████████▎        | 90/120 [13:16<04:03,  8.13s/it]
                                                                         
{'loss': 5.0231, 'grad_norm': 1.1817212104797363, 'learning_rate': 3.6698754619112974e-06, 'ppl': 151.88141, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 216.71510314941406, 'tokens/total': 1449984, 'tokens/trainable': 656140, 'epoch': 6.94}
-
 75%|██████████████████████████▎        | 90/120 [13:16<04:03,  8.13s/it]
 76%|██████████████████████████▌        | 91/120 [13:22<03:39,  7.57s/it]
 77%|██████████████████████████▊        | 92/120 [13:32<03:54,  8.36s/it]
 78%|███████████████████████████▏       | 93/120 [13:39<03:37,  8.05s/it]
 78%|███████████████████████████▍       | 94/120 [13:48<03:30,  8.11s/it]
 79%|███████████████████████████▋       | 95/120 [13:56<03:23,  8.15s/it]
                                                                         
{'loss': 4.9815, 'grad_norm': 1.2491371631622314, 'learning_rate': 2.6325886212359496e-06, 'ppl': 145.69276, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 233.44076538085938, 'tokens/total': 1527808, 'tokens/trainable': 690609, 'epoch': 7.31}
-
 79%|███████████████████████████▋       | 95/120 [13:56<03:23,  8.15s/it]
 80%|████████████████████████████       | 96/120 [14:04<03:16,  8.18s/it]
 81%|████████████████████████████▎      | 97/120 [14:12<03:08,  8.20s/it]
 82%|████████████████████████████▌      | 98/120 [14:21<03:00,  8.22s/it]
 82%|████████████████████████████▉      | 99/120 [14:29<02:52,  8.23s/it]
 83%|████████████████████████████▎     | 100/120 [14:36<02:39,  7.96s/it]
                                                                         
{'loss': 4.9679, 'grad_norm': 1.2302298545837402, 'learning_rate': 1.74528103037226e-06, 'ppl': 143.72475, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 247.9733123779297, 'tokens/total': 1609728, 'tokens/trainable': 727991, 'epoch': 7.71}
-
 83%|████████████████████████████▎     | 100/120 [14:36<02:39,  7.96s/it][2026-01-24 13:07:30,848] [INFO] [axolotl.core.trainers.base.evaluate:400] [PID:6937] Running evaluation step...
+
                                                                         
  0%|                                 | 1/650 [01:05<11:52:51, 65.90s/it]
  0%|                                  | 2/650 [01:14<5:45:12, 31.96s/it]
  0%|▏                                 | 3/650 [01:22<3:47:42, 21.12s/it]
  1%|▏                                 | 4/650 [01:30<2:52:35, 16.03s/it]
  1%|▎                                 | 5/650 [01:38<2:22:09, 13.22s/it]
                                                                         
{'loss': 6.0638, 'grad_norm': 0.5395623445510864, 'learning_rate': 4.000000000000001e-06, 'ppl': 430.00636, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 215.98207092285156, 'tokens/total': 81920, 'tokens/trainable': 36915, 'epoch': 0.39}
+
  1%|▎                                 | 5/650 [01:38<2:22:09, 13.22s/it]
  1%|▎                                 | 6/650 [01:47<2:03:51, 11.54s/it]
  1%|▎                                 | 7/650 [01:55<1:52:10, 10.47s/it]
  1%|▍                                 | 8/650 [02:03<1:44:28,  9.76s/it]
  1%|▍                                 | 9/650 [02:11<1:39:17,  9.29s/it]
  2%|▌                                | 10/650 [02:20<1:35:45,  8.98s/it]
                                                                         
{'loss': 6.0041, 'grad_norm': 0.5103967785835266, 'learning_rate': 9e-06, 'ppl': 405.08625, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 218.7120819091797, 'tokens/total': 163840, 'tokens/trainable': 74494, 'epoch': 0.78}
+
  2%|▌                                | 10/650 [02:20<1:35:45,  8.98s/it]
  2%|▌                                | 11/650 [02:28<1:33:17,  8.76s/it]
  2%|▌                                | 12/650 [02:36<1:31:29,  8.60s/it]
  2%|▋                                | 13/650 [02:42<1:23:44,  7.89s/it]
  2%|▋                                | 14/650 [02:52<1:30:40,  8.55s/it]
  2%|▊                                | 15/650 [03:01<1:29:34,  8.46s/it]
                                                                         
{'loss': 6.0541, 'grad_norm': 0.6589650511741638, 'learning_rate': 1.4e-05, 'ppl': 425.85546, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 198.2006378173828, 'tokens/total': 241664, 'tokens/trainable': 109022, 'epoch': 1.16}
+
  2%|▊                                | 15/650 [03:01<1:29:34,  8.46s/it]
  2%|▊                                | 16/650 [03:09<1:28:47,  8.40s/it]
  3%|▊                                | 17/650 [03:17<1:28:11,  8.36s/it]
  3%|▉                                | 18/650 [03:26<1:27:44,  8.33s/it]
  3%|▉                                | 19/650 [03:34<1:27:22,  8.31s/it]
  3%|█                                | 20/650 [03:42<1:27:04,  8.29s/it]
                                                                         
{'loss': 6.0239, 'grad_norm': 0.7923386096954346, 'learning_rate': 1.9e-05, 'ppl': 413.18689, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 237.59417724609375, 'tokens/total': 323584, 'tokens/trainable': 146859, 'epoch': 1.55}
+
  3%|█                                | 20/650 [03:42<1:27:04,  8.29s/it]
  3%|█                                | 21/650 [03:50<1:26:50,  8.28s/it]
  3%|█                                | 22/650 [03:59<1:26:34,  8.27s/it]
  4%|█▏                               | 23/650 [04:07<1:26:24,  8.27s/it]
  4%|█▏                               | 24/650 [04:15<1:26:14,  8.27s/it]
  4%|█▎                               | 25/650 [04:23<1:26:06,  8.27s/it]
                                                                         
{'loss': 5.9298, 'grad_norm': 0.9688937664031982, 'learning_rate': 1.9998010727705237e-05, 'ppl': 376.07929, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 212.3710479736328, 'tokens/total': 405504, 'tokens/trainable': 183924, 'epoch': 1.94}
+
  4%|█▎                               | 25/650 [04:23<1:26:06,  8.27s/it]
  4%|█▎                               | 26/650 [04:30<1:19:40,  7.66s/it]
  4%|█▎                               | 27/650 [04:40<1:27:26,  8.42s/it]
  4%|█▍                               | 28/650 [04:48<1:26:49,  8.38s/it]
  4%|█▍                               | 29/650 [04:56<1:26:19,  8.34s/it]
  5%|█▌                               | 30/650 [05:05<1:25:56,  8.32s/it]
                                                                         
{'loss': 5.8764, 'grad_norm': 1.120764136314392, 'learning_rate': 1.9989930665413148e-05, 'ppl': 356.52344, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 214.12718200683594, 'tokens/total': 483328, 'tokens/trainable': 218753, 'epoch': 2.31}
+
  5%|█▌                               | 30/650 [05:05<1:25:56,  8.32s/it]
  5%|█▌                               | 31/650 [05:13<1:25:38,  8.30s/it]
  5%|█▌                               | 32/650 [05:21<1:25:24,  8.29s/it]
  5%|█▋                               | 33/650 [05:29<1:25:11,  8.28s/it]
  5%|█▋                               | 34/650 [05:38<1:24:59,  8.28s/it]
  5%|█▊                               | 35/650 [05:46<1:24:45,  8.27s/it]
                                                                         
{'loss': 5.8105, 'grad_norm': 1.2133527994155884, 'learning_rate': 1.9975640502598243e-05, 'ppl': 333.78598, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 220.15711975097656, 'tokens/total': 565248, 'tokens/trainable': 256149, 'epoch': 2.71}
+
  5%|█▊                               | 35/650 [05:46<1:24:45,  8.27s/it]
  6%|█▊                               | 36/650 [05:54<1:24:34,  8.27s/it]
  6%|█▉                               | 37/650 [06:02<1:24:24,  8.26s/it]
  6%|█▉                               | 38/650 [06:11<1:24:15,  8.26s/it]
  6%|█▉                               | 39/650 [06:17<1:17:55,  7.65s/it]
  6%|██                               | 40/650 [06:27<1:25:27,  8.41s/it]
                                                                         
{'loss': 5.6921, 'grad_norm': 1.2973392009735107, 'learning_rate': 1.995514912254015e-05, 'ppl': 296.51565, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 220.37948608398438, 'tokens/total': 643072, 'tokens/trainable': 290596, 'epoch': 3.08}
+
  6%|██                               | 40/650 [06:27<1:25:27,  8.41s/it]
  6%|██                               | 41/650 [06:35<1:24:52,  8.36s/it]
  6%|██▏                              | 42/650 [06:44<1:24:24,  8.33s/it]
  7%|██▏                              | 43/650 [06:52<1:24:02,  8.31s/it]
  7%|██▏                              | 44/650 [07:00<1:23:44,  8.29s/it]
  7%|██▎                              | 45/650 [07:08<1:23:30,  8.28s/it]
                                                                         
{'loss': 5.5982, 'grad_norm': 1.3150535821914673, 'learning_rate': 1.9928469263418376e-05, 'ppl': 269.94008, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 223.76809692382812, 'tokens/total': 724992, 'tokens/trainable': 328311, 'epoch': 3.47}
+
  7%|██▎                              | 45/650 [07:08<1:23:30,  8.28s/it]
  7%|██▎                              | 46/650 [07:17<1:23:18,  8.28s/it]
  7%|██▍                              | 47/650 [07:25<1:23:05,  8.27s/it]
  7%|██▍                              | 48/650 [07:33<1:22:54,  8.26s/it]
  8%|██▍                              | 49/650 [07:41<1:22:43,  8.26s/it]
  8%|██▌                              | 50/650 [07:50<1:22:33,  8.26s/it]
                                                                         
{'loss': 5.394, 'grad_norm': 1.2739847898483276, 'learning_rate': 1.9895617510393773e-05, 'ppl': 220.08196, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.82, 'tokens/train_per_sec_per_gpu': 209.72315979003906, 'tokens/total': 806912, 'tokens/trainable': 365622, 'epoch': 3.86}
+
  8%|██▌                              | 50/650 [07:50<1:22:33,  8.26s/it][2026-01-24 13:33:23,098] [INFO] [axolotl.core.trainers.base.evaluate:400] [PID:9359] Running evaluation step...
 
 
  0%|                                            | 0/100 [00:00<?, ?it/s]
-
  2%|▋                                   | 2/100 [00:00<00:25,  3.84it/s]
-
  3%|█                                   | 3/100 [00:01<00:35,  2.70it/s]
-
  4%|█▍                                  | 4/100 [00:01<00:41,  2.33it/s]
-
  5%|█▊                                  | 5/100 [00:02<00:54,  1.74it/s]
-
  6%|██▏                                 | 6/100 [00:02<00:52,  1.79it/s]
-
  7%|██▌                                 | 7/100 [00:03<00:50,  1.83it/s]
-
  8%|██▉                                 | 8/100 [00:03<00:49,  1.85it/s]
-
  9%|███▏                                | 9/100 [00:04<00:49,  1.83it/s]
+
  2%|▋                                   | 2/100 [00:00<00:25,  3.85it/s]
+
  3%|█                                   | 3/100 [00:01<00:35,  2.71it/s]
+
  4%|█▍                                  | 4/100 [00:01<00:40,  2.35it/s]
+
  5%|█▊                                  | 5/100 [00:02<00:54,  1.75it/s]
+
  6%|██▏                                 | 6/100 [00:02<00:52,  1.80it/s]
+
  7%|██▌                                 | 7/100 [00:03<00:50,  1.84it/s]
+
  8%|██▉                                 | 8/100 [00:03<00:49,  1.86it/s]
+
  9%|███▏                                | 9/100 [00:04<00:49,  1.84it/s]
+
 10%|███▌                               | 10/100 [00:05<00:48,  1.87it/s]
+
 11%|███▊                               | 11/100 [00:05<00:47,  1.88it/s]
+
 12%|████▏                              | 12/100 [00:06<00:46,  1.89it/s]
+
 13%|████▌                              | 13/100 [00:06<00:46,  1.86it/s]
+
 14%|████▉                              | 14/100 [00:07<00:45,  1.88it/s]
+
 15%|█████▎                             | 15/100 [00:07<00:44,  1.89it/s]
+
 16%|█████▌                             | 16/100 [00:08<00:44,  1.90it/s]
+
 17%|█████▉                             | 17/100 [00:08<00:44,  1.86it/s]
+
 18%|██████▎                            | 18/100 [00:09<00:43,  1.88it/s]
+
 19%|██████▋                            | 19/100 [00:09<00:42,  1.89it/s]
+
 20%|███████                            | 20/100 [00:10<00:42,  1.90it/s]
+
 21%|███████▎                           | 21/100 [00:10<00:42,  1.86it/s]
+
 22%|███████▋                           | 22/100 [00:11<00:41,  1.88it/s]
+
 23%|████████                           | 23/100 [00:11<00:40,  1.89it/s]
+
 24%|████████▍                          | 24/100 [00:12<00:40,  1.90it/s]
+
 25%|████████▊                          | 25/100 [00:13<00:40,  1.86it/s]
+
 26%|█████████                          | 26/100 [00:13<00:39,  1.88it/s]
+
 27%|█████████▍                         | 27/100 [00:14<00:38,  1.89it/s]
+
 28%|█████████▊                         | 28/100 [00:14<00:37,  1.90it/s]
+
 29%|██████████▏                        | 29/100 [00:15<00:38,  1.86it/s]
+
 30%|██████████▌                        | 30/100 [00:15<00:37,  1.88it/s]
+
 31%|██████████▊                        | 31/100 [00:16<00:36,  1.89it/s]
+
 32%|███████████▏                       | 32/100 [00:16<00:35,  1.90it/s]
+
 33%|███████████▌                       | 33/100 [00:17<00:35,  1.86it/s]
+
 34%|███████████▉                       | 34/100 [00:17<00:35,  1.88it/s]
+
 35%|████████████▎                      | 35/100 [00:18<00:34,  1.89it/s]
+
 36%|████████████▌                      | 36/100 [00:18<00:33,  1.90it/s]
+
 37%|████████████▉                      | 37/100 [00:19<00:33,  1.86it/s]
+
 38%|█████████████▎                     | 38/100 [00:19<00:32,  1.88it/s]
+
 39%|█████████████▋                     | 39/100 [00:20<00:32,  1.89it/s]
+
 40%|██████████████                     | 40/100 [00:20<00:31,  1.90it/s]
+
 41%|██████████████▎                    | 41/100 [00:21<00:31,  1.86it/s]
+
 42%|██████████████▋                    | 42/100 [00:22<00:30,  1.88it/s]
+
 43%|███████████████                    | 43/100 [00:22<00:30,  1.89it/s]
+
 44%|███████████████▍                   | 44/100 [00:23<00:29,  1.90it/s]
+
 45%|███████████████▊                   | 45/100 [00:23<00:29,  1.86it/s]
+
 46%|████████████████                   | 46/100 [00:24<00:28,  1.88it/s]
+
 47%|████████████████▍                  | 47/100 [00:24<00:28,  1.89it/s]
+
 48%|████████████████▊                  | 48/100 [00:25<00:27,  1.90it/s]
+
 49%|█████████████████▏                 | 49/100 [00:25<00:27,  1.86it/s]
+
 50%|█████████████████▌                 | 50/100 [00:26<00:26,  1.88it/s]
+
 51%|█████████████████▊                 | 51/100 [00:26<00:25,  1.89it/s]
+
 52%|██████████████████▏                | 52/100 [00:27<00:25,  1.90it/s]
+
 53%|██████████████████▌                | 53/100 [00:27<00:25,  1.86it/s]
+
 54%|██████████████████▉                | 54/100 [00:28<00:24,  1.88it/s]
+
 55%|███████████████████▎               | 55/100 [00:28<00:23,  1.89it/s]
+
 56%|███████████████████▌               | 56/100 [00:29<00:23,  1.90it/s]
+
 57%|███████████████████▉               | 57/100 [00:30<00:23,  1.86it/s]
+
 58%|████████████████████▎              | 58/100 [00:30<00:22,  1.88it/s]
+
 59%|████████████████████▋              | 59/100 [00:31<00:21,  1.89it/s]
+
 60%|█████████████████████              | 60/100 [00:31<00:21,  1.90it/s]
+
 61%|█████████████████████▎             | 61/100 [00:32<00:20,  1.86it/s]
+
 62%|█████████████████████▋             | 62/100 [00:32<00:20,  1.88it/s]
+
 63%|██████████████████████             | 63/100 [00:33<00:19,  1.89it/s]
+
 64%|██████████████████████▍            | 64/100 [00:33<00:18,  1.90it/s]
+
 65%|██████████████████████▊            | 65/100 [00:34<00:18,  1.86it/s]
+
 66%|███████████████████████            | 66/100 [00:34<00:18,  1.88it/s]
+
 67%|███████████████████████▍           | 67/100 [00:35<00:17,  1.89it/s]
+
 68%|███████████████████████▊           | 68/100 [00:35<00:16,  1.90it/s]
+
 69%|████████████████████████▏          | 69/100 [00:36<00:16,  1.86it/s]
+
 70%|████████████████████████▌          | 70/100 [00:36<00:15,  1.88it/s]
+
 71%|████████████████████████▊          | 71/100 [00:37<00:15,  1.89it/s]
+
 72%|█████████████████████████▏         | 72/100 [00:37<00:14,  1.90it/s]
+
 73%|█████████████████████████▌         | 73/100 [00:38<00:14,  1.86it/s]
+
 74%|█████████████████████████▉         | 74/100 [00:39<00:13,  1.88it/s]
+
 75%|██████████████████████████▎        | 75/100 [00:39<00:13,  1.89it/s]
+
 76%|██████████████████████████▌        | 76/100 [00:40<00:12,  1.90it/s]
+
 77%|██████████████████████████▉        | 77/100 [00:40<00:12,  1.86it/s]
+
 78%|███████████████████████████▎       | 78/100 [00:41<00:11,  1.88it/s]
+
 79%|███████████████████████████▋       | 79/100 [00:41<00:11,  1.89it/s]
+
 80%|████████████████████████████       | 80/100 [00:42<00:10,  1.90it/s]
+
 81%|████████████████████████████▎      | 81/100 [00:42<00:10,  1.86it/s]
+
 82%|████████████████████████████▋      | 82/100 [00:43<00:09,  1.88it/s]
+
 83%|█████████████████████████████      | 83/100 [00:43<00:08,  1.89it/s]
+
 84%|█████████████████████████████▍     | 84/100 [00:44<00:08,  1.90it/s]
+
 85%|█████████████████████████████▊     | 85/100 [00:44<00:08,  1.86it/s]
+
 86%|██████████████████████████████     | 86/100 [00:45<00:07,  1.88it/s]
+
 87%|██████████████████████████████▍    | 87/100 [00:45<00:06,  1.89it/s]
+
 88%|██████████████████████████████▊    | 88/100 [00:46<00:06,  1.90it/s]
+
 89%|███████████████████████████████▏   | 89/100 [00:46<00:05,  1.86it/s]
+
 90%|███████████████████████████████▌   | 90/100 [00:47<00:05,  1.88it/s]
+
 91%|███████████████████████████████▊   | 91/100 [00:48<00:04,  1.89it/s]
+
 92%|████████████████████████████████▏  | 92/100 [00:48<00:04,  1.90it/s]
+
 93%|████████████████████████████████▌  | 93/100 [00:49<00:03,  1.86it/s]
+
 94%|████████████████████████████████▉  | 94/100 [00:49<00:03,  1.88it/s]
+
 95%|█████████████████████████████████▎ | 95/100 [00:50<00:02,  1.89it/s]
+
 96%|█████████████████████████████████▌ | 96/100 [00:50<00:02,  1.90it/s]
+
 97%|█████████████████████████████████▉ | 97/100 [00:51<00:01,  1.86it/s]
+
 98%|██████████████████████████████████▎| 98/100 [00:51<00:01,  1.88it/s]
+
 99%|██████████████████████████████████▋| 99/100 [00:52<00:00,  1.89it/s]
+
100%|██████████████████████████████████| 100/100 [00:52<00:00,  1.83it/s]
                                                                         
+
                                                                         
{'eval_loss': 2.1003942489624023, 'eval_runtime': 53.9477, 'eval_samples_per_second': 3.707, 'eval_steps_per_second': 1.854, 'eval_ppl': 8.16939, 'memory/max_active (GiB)': 14.84, 'memory/max_allocated (GiB)': 14.84, 'memory/device_reserved (GiB)': 31.82, 'epoch': 3.86, 'tokens/train_per_sec_per_gpu': 0.0}
+
  8%|██▌                              | 50/650 [08:44<1:22:33,  8.26s/it]
+
100%|██████████████████████████████████| 100/100 [00:53<00:00,  1.83it/s]
+
                                                                         
  8%|██▌                              | 51/650 [08:52<4:04:05, 24.45s/it]
  8%|██▋                              | 52/650 [08:58<3:09:14, 18.99s/it]
  8%|██▋                              | 53/650 [09:08<2:42:38, 16.35s/it]
  8%|██▋                              | 54/650 [09:16<2:15:29, 13.64s/it]
  8%|██▊                              | 55/650 [09:24<1:59:13, 12.02s/it]
                                                                         
{'loss': 5.3139, 'grad_norm': 1.290380597114563, 'learning_rate': 1.985661428529863e-05, 'ppl': 203.14094, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 238.74021911621094, 'tokens/total': 884736, 'tokens/trainable': 400183, 'epoch': 4.24}
+
  8%|██▊                              | 55/650 [09:24<1:59:13, 12.02s/it]
  9%|██▊                              | 56/650 [09:32<1:47:47, 10.89s/it]
  9%|██▉                              | 57/650 [09:40<1:39:49, 10.10s/it]
  9%|██▉                              | 58/650 [09:48<1:31:27,  9.27s/it]
  9%|██▉                              | 59/650 [09:56<1:28:15,  8.96s/it]
  9%|███                              | 60/650 [10:04<1:25:58,  8.74s/it]
                                                                         
{'loss': 5.2056, 'grad_norm': 1.3334054946899414, 'learning_rate': 1.9811483833941726e-05, 'ppl': 182.29021, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 226.33343505859375, 'tokens/total': 966656, 'tokens/trainable': 437640, 'epoch': 4.63}
+
  9%|███                              | 60/650 [10:04<1:25:58,  8.74s/it]
  9%|███                              | 61/650 [10:12<1:24:22,  8.59s/it]
 10%|███▏                             | 62/650 [10:21<1:23:11,  8.49s/it]
 10%|███▏                             | 63/650 [10:29<1:22:19,  8.41s/it]
 10%|███▏                             | 64/650 [10:37<1:21:39,  8.36s/it]
 10%|███▎                             | 65/650 [10:43<1:15:16,  7.72s/it]
                                                                         
{'loss': 5.0582, 'grad_norm': 1.161880612373352, 'learning_rate': 1.9760254211036245e-05, 'ppl': 157.30711, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 187.5677032470703, 'tokens/total': 1044480, 'tokens/trainable': 472345, 'epoch': 5.0}
+
 10%|███▎                             | 65/650 [10:43<1:15:16,  7.72s/it]
 10%|███▎                             | 66/650 [10:53<1:22:08,  8.44s/it]
 10%|███▍                             | 67/650 [11:02<1:21:26,  8.38s/it]
 10%|███▍                             | 68/650 [11:10<1:20:55,  8.34s/it]
 11%|███▌                             | 69/650 [11:18<1:20:30,  8.31s/it]
 11%|███▌                             | 70/650 [11:26<1:20:12,  8.30s/it]
                                                                         
{'loss': 4.9713, 'grad_norm': 1.14654541015625, 'learning_rate': 1.9702957262759964e-05, 'ppl': 144.21424, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 228.5047607421875, 'tokens/total': 1126400, 'tokens/trainable': 509679, 'epoch': 5.39}
+
 11%|███▌                             | 70/650 [11:26<1:20:12,  8.30s/it]
 11%|███▌                             | 71/650 [11:35<1:19:56,  8.28s/it]
 11%|███▋                             | 72/650 [11:43<1:19:42,  8.27s/it]
 11%|███▋                             | 73/650 [11:51<1:19:28,  8.27s/it]
 11%|███▊                             | 74/650 [11:59<1:19:17,  8.26s/it]
 12%|███▊                             | 75/650 [12:08<1:19:07,  8.26s/it]
                                                                         
{'loss': 4.8927, 'grad_norm': 1.1394518613815308, 'learning_rate': 1.9639628606958535e-05, 'ppl': 133.31303, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 227.03880310058594, 'tokens/total': 1208320, 'tokens/trainable': 546748, 'epoch': 5.78}
+
 12%|███▊                             | 75/650 [12:08<1:19:07,  8.26s/it]
 12%|███▊                             | 76/650 [12:16<1:18:58,  8.26s/it]
 12%|███▉                             | 77/650 [12:24<1:18:51,  8.26s/it]
 12%|███▉                             | 78/650 [12:30<1:12:58,  7.66s/it]
 12%|████                             | 79/650 [12:40<1:19:33,  8.36s/it]
 12%|████                             | 80/650 [12:49<1:19:05,  8.33s/it]
                                                                         
{'loss': 4.8296, 'grad_norm': 1.182919979095459, 'learning_rate': 1.9570307611004124e-05, 'ppl': 125.16089, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 229.55503845214844, 'tokens/total': 1286144, 'tokens/trainable': 581786, 'epoch': 6.16}
+
 12%|████                             | 80/650 [12:49<1:19:05,  8.33s/it]
 12%|████                             | 81/650 [12:57<1:18:45,  8.31s/it]
 13%|████▏                            | 82/650 [13:05<1:18:28,  8.29s/it]
 13%|████▏                            | 83/650 [13:13<1:18:13,  8.28s/it]
 13%|████▎                            | 84/650 [13:22<1:18:00,  8.27s/it]
 13%|████▎                            | 85/650 [13:30<1:17:50,  8.27s/it]
                                                                         
{'loss': 4.7205, 'grad_norm': 1.1635547876358032, 'learning_rate': 1.9495037367323264e-05, 'ppl': 112.22435, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 238.2696990966797, 'tokens/total': 1368064, 'tokens/trainable': 619252, 'epoch': 6.55}
+
 13%|████▎                            | 85/650 [13:30<1:17:50,  8.27s/it]
 13%|████▎                            | 86/650 [13:38<1:17:42,  8.27s/it]
 13%|████▍                            | 87/650 [13:46<1:17:33,  8.26s/it]
 14%|████▍                            | 88/650 [13:55<1:17:23,  8.26s/it]
 14%|████▌                            | 89/650 [14:03<1:17:15,  8.26s/it]
 14%|████▌                            | 90/650 [14:11<1:17:04,  8.26s/it]
                                                                         
{'loss': 4.6447, 'grad_norm': 1.1828676462173462, 'learning_rate': 1.9413864666609036e-05, 'ppl': 104.03215, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 216.6095428466797, 'tokens/total': 1449984, 'tokens/trainable': 656250, 'epoch': 6.94}
+
 14%|████▌                            | 90/650 [14:11<1:17:04,  8.26s/it]
 14%|████▌                            | 91/650 [14:18<1:11:20,  7.66s/it]
 14%|████▋                            | 92/650 [14:28<1:18:02,  8.39s/it]
 14%|████▋                            | 93/650 [14:36<1:17:33,  8.35s/it]
 14%|████▊                            | 94/650 [14:44<1:17:10,  8.33s/it]
 15%|████▊                            | 95/650 [14:52<1:16:51,  8.31s/it]
                                                                         
{'loss': 4.6145, 'grad_norm': 1.196416974067688, 'learning_rate': 1.9326839968734278e-05, 'ppl': 100.93735, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 227.1728973388672, 'tokens/total': 1527808, 'tokens/trainable': 691447, 'epoch': 7.31}
+
 15%|████▊                            | 95/650 [14:52<1:16:51,  8.31s/it]
 15%|████▊                            | 96/650 [15:01<1:16:35,  8.30s/it]
 15%|████▉                            | 97/650 [15:09<1:16:20,  8.28s/it]
 15%|████▉                            | 98/650 [15:17<1:16:06,  8.27s/it]
 15%|█████                            | 99/650 [15:25<1:15:56,  8.27s/it]
 15%|████▉                           | 100/650 [15:34<1:15:46,  8.27s/it]
                                                                         
{'loss': 4.4484, 'grad_norm': 1.2008439302444458, 'learning_rate': 1.9234017371383946e-05, 'ppl': 85.49005, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 229.3255615234375, 'tokens/total': 1609728, 'tokens/trainable': 728444, 'epoch': 7.71}
+
 15%|████▉                           | 100/650 [15:34<1:15:46,  8.27s/it][2026-01-24 13:41:07,232] [INFO] [axolotl.core.trainers.base.evaluate:400] [PID:9359] Running evaluation step...
+
+
  0%|                                            | 0/100 [00:00<?, ?it/s]
+
  2%|▋                                   | 2/100 [00:00<00:25,  3.85it/s]
+
  3%|█                                   | 3/100 [00:01<00:35,  2.71it/s]
+
  4%|█▍                                  | 4/100 [00:01<00:40,  2.35it/s]
+
  5%|█▊                                  | 5/100 [00:02<00:54,  1.75it/s]
+
  6%|██▏                                 | 6/100 [00:02<00:52,  1.80it/s]
+
  7%|██▌                                 | 7/100 [00:03<00:50,  1.84it/s]
+
  8%|██▉                                 | 8/100 [00:03<00:49,  1.86it/s]
+
  9%|███▏                                | 9/100 [00:04<00:49,  1.84it/s]
 
 10%|███▌                               | 10/100 [00:05<00:48,  1.86it/s]
-
 11%|███▊                               | 11/100 [00:05<00:47,  1.87it/s]
+
 11%|███▊                               | 11/100 [00:05<00:47,  1.88it/s]
 
 12%|████▏                              | 12/100 [00:06<00:46,  1.89it/s]
-
 13%|████▌                              | 13/100 [00:06<00:46,  1.85it/s]
-
 14%|████▉                              | 14/100 [00:07<00:45,  1.87it/s]
-
 15%|█████▎                             | 15/100 [00:07<00:45,  1.88it/s]
-
 16%|█████▌                             | 16/100 [00:08<00:44,  1.89it/s]
+
 13%|████▌                              | 13/100 [00:06<00:46,  1.86it/s]
+
 14%|████▉                              | 14/100 [00:07<00:45,  1.88it/s]
+
 15%|█████▎                             | 15/100 [00:07<00:44,  1.89it/s]
+
 16%|█████▌                             | 16/100 [00:08<00:44,  1.90it/s]
 
 17%|█████▉                             | 17/100 [00:08<00:44,  1.86it/s]
 
 18%|██████▎                            | 18/100 [00:09<00:43,  1.88it/s]
 
 19%|██████▋                            | 19/100 [00:09<00:42,  1.89it/s]
-
 20%|███████                            | 20/100 [00:10<00:42,  1.89it/s]
+
 20%|███████                            | 20/100 [00:10<00:42,  1.90it/s]
 
 21%|███████▎                           | 21/100 [00:10<00:42,  1.86it/s]
 
 22%|███████▋                           | 22/100 [00:11<00:41,  1.88it/s]
 
 23%|████████                           | 23/100 [00:11<00:40,  1.89it/s]
@@ -396,14 +502,14 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 30%|██████████▌                        | 30/100 [00:15<00:37,  1.88it/s]
 
 31%|██████████▊                        | 31/100 [00:16<00:36,  1.89it/s]
 
 32%|███████████▏                       | 32/100 [00:16<00:35,  1.90it/s]
-
 33%|███████████▌                       | 33/100 [00:17<00:36,  1.86it/s]
+
 33%|███████████▌                       | 33/100 [00:17<00:35,  1.86it/s]
 
 34%|███████████▉                       | 34/100 [00:17<00:35,  1.88it/s]
 
 35%|████████████▎                      | 35/100 [00:18<00:34,  1.89it/s]
 
 36%|████████████▌                      | 36/100 [00:18<00:33,  1.90it/s]
 
 37%|████████████▉                      | 37/100 [00:19<00:33,  1.86it/s]
 
 38%|█████████████▎                     | 38/100 [00:19<00:32,  1.88it/s]
 
 39%|█████████████▋                     | 39/100 [00:20<00:32,  1.89it/s]
-
 40%|██████████████                     | 40/100 [00:21<00:31,  1.90it/s]
+
 40%|██████████████                     | 40/100 [00:20<00:31,  1.90it/s]
 
 41%|██████████████▎                    | 41/100 [00:21<00:31,  1.86it/s]
 
 42%|██████████████▋                    | 42/100 [00:22<00:30,  1.88it/s]
 
 43%|███████████████                    | 43/100 [00:22<00:30,  1.89it/s]
@@ -415,15 +521,15 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 49%|█████████████████▏                 | 49/100 [00:25<00:27,  1.86it/s]
 
 50%|█████████████████▌                 | 50/100 [00:26<00:26,  1.88it/s]
 
 51%|█████████████████▊                 | 51/100 [00:26<00:25,  1.89it/s]
-
 52%|██████████████████▏                | 52/100 [00:27<00:25,  1.89it/s]
+
 52%|██████████████████▏                | 52/100 [00:27<00:25,  1.90it/s]
 
 53%|██████████████████▌                | 53/100 [00:27<00:25,  1.86it/s]
 
 54%|██████████████████▉                | 54/100 [00:28<00:24,  1.88it/s]
 
 55%|███████████████████▎               | 55/100 [00:28<00:23,  1.89it/s]
-
 56%|███████████████████▌               | 56/100 [00:29<00:23,  1.89it/s]
+
 56%|███████████████████▌               | 56/100 [00:29<00:23,  1.90it/s]
 
 57%|███████████████████▉               | 57/100 [00:30<00:23,  1.86it/s]
 
 58%|████████████████████▎              | 58/100 [00:30<00:22,  1.88it/s]
 
 59%|████████████████████▋              | 59/100 [00:31<00:21,  1.89it/s]
-
 60%|█████████████████████              | 60/100 [00:31<00:21,  1.89it/s]
+
 60%|█████████████████████              | 60/100 [00:31<00:21,  1.90it/s]
 
 61%|█████████████████████▎             | 61/100 [00:32<00:20,  1.86it/s]
 
 62%|█████████████████████▋             | 62/100 [00:32<00:20,  1.88it/s]
 
 63%|██████████████████████             | 63/100 [00:33<00:19,  1.89it/s]
@@ -435,22 +541,22 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 69%|████████████████████████▏          | 69/100 [00:36<00:16,  1.86it/s]
 
 70%|████��███████████████████▌          | 70/100 [00:36<00:15,  1.88it/s]
 
 71%|████████████████████████▊          | 71/100 [00:37<00:15,  1.89it/s]
-
 72%|█████████████████████████▏         | 72/100 [00:38<00:14,  1.90it/s]
+
 72%|█████████████████████████▏         | 72/100 [00:37<00:14,  1.90it/s]
 
 73%|█████████████████████████▌         | 73/100 [00:38<00:14,  1.86it/s]
 
 74%|█████████████████████████▉         | 74/100 [00:39<00:13,  1.88it/s]
 
 75%|██████████████████████████▎        | 75/100 [00:39<00:13,  1.89it/s]
-
 76%|██████████████████████████▌        | 76/100 [00:40<00:12,  1.89it/s]
+
 76%|██████████████████████████▌        | 76/100 [00:40<00:12,  1.90it/s]
 
 77%|██████████████████████████▉        | 77/100 [00:40<00:12,  1.86it/s]
 
 78%|███████████████████████████▎       | 78/100 [00:41<00:11,  1.88it/s]
 
 79%|███████████████████████████▋       | 79/100 [00:41<00:11,  1.89it/s]
-
 80%|████████████████████████████       | 80/100 [00:42<00:10,  1.89it/s]
+
 80%|████████████████████████████       | 80/100 [00:42<00:10,  1.90it/s]
 
 81%|████████████████████████████▎      | 81/100 [00:42<00:10,  1.86it/s]
 
 82%|████████████████████████████▋      | 82/100 [00:43<00:09,  1.88it/s]
 
 83%|█████████████████████████████      | 83/100 [00:43<00:08,  1.89it/s]
 
 84%|█████████████████████████████▍     | 84/100 [00:44<00:08,  1.90it/s]
 
 85%|█████████████████████████████▊     | 85/100 [00:44<00:08,  1.86it/s]
 
 86%|██████████████████████████████     | 86/100 [00:45<00:07,  1.88it/s]
-
 87%|██████████████████████████████▍    | 87/100 [00:46<00:06,  1.89it/s]
+
 87%|██████████████████████████████▍    | 87/100 [00:45<00:06,  1.89it/s]
 
 88%|██████████████████████████████▊    | 88/100 [00:46<00:06,  1.90it/s]
 
 89%|███████████████████████████████▏   | 89/100 [00:47<00:05,  1.86it/s]
 
 90%|███████████████████████████████▌   | 90/100 [00:47<00:05,  1.88it/s]
@@ -464,58 +570,7 @@ trainable params: 1,572,864 || all params: 3,837,594,624 || trainable%: 0.0410
 
 98%|██████████████████████████████████▎| 98/100 [00:51<00:01,  1.88it/s]
 
 99%|██████████████████████████████████▋| 99/100 [00:52<00:00,  1.89it/s]
 
100%|██████████████████████████████████| 100/100 [00:52<00:00,  1.83it/s]
                                                                         
-
                                                                         
{'eval_loss': 2.0937912464141846, 'eval_runtime': 54.204, 'eval_samples_per_second': 3.69, 'eval_steps_per_second': 1.845, 'eval_ppl': 8.11563, 'memory/max_active (GiB)': 14.84, 'memory/max_allocated (GiB)': 14.84, 'memory/device_reserved (GiB)': 31.82, 'epoch': 7.71, 'tokens/train_per_sec_per_gpu': 0.0}
-
 83%|████████████████████████████▎     | 100/120 [15:30<02:39,  7.96s/it]
+
                                                                         
{'eval_loss': 2.0366907119750977, 'eval_runtime': 53.964, 'eval_samples_per_second': 3.706, 'eval_steps_per_second': 1.853, 'eval_ppl': 7.6652, 'memory/max_active (GiB)': 14.84, 'memory/max_allocated (GiB)': 14.84, 'memory/device_reserved (GiB)': 31.84, 'epoch': 7.71, 'tokens/train_per_sec_per_gpu': 0.0}
+
 15%|████▉                           | 100/650 [16:28<1:15:46,  8.27s/it]
 
100%|██████████████████████████████████| 100/100 [00:53<00:00,  1.83it/s]
-
                                                                         
 84%|████████████████████████████▌     | 101/120 [15:39<07:42, 24.33s/it]
 85%|████████████████████████████▉     | 102/120 [15:47<05:51, 19.51s/it]
 86%|█████████████████████████████▏    | 103/120 [15:55<04:34, 16.14s/it]
 87%|█████████████████████████████▍    | 104/120 [16:02<03:30, 13.17s/it]
 88%|█████████████████████████████▊    | 105/120 [16:12<03:04, 12.32s/it]
                                                                         
{'loss': 4.9415, 'grad_norm': 1.301143765449524, 'learning_rate': 1.0260157130864178e-06, 'ppl': 139.98006, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 209.69256591796875, 'tokens/total': 1687552, 'tokens/trainable': 763004, 'epoch': 8.08}
-
 88%|█████████████████████████████▊    | 105/120 [16:12<03:04, 12.32s/it]
 88%|██████████████████████████████    | 106/120 [16:20<02:35, 11.11s/it]
 89%|██████████████████████████████▎   | 107/120 [16:28<02:13, 10.26s/it]
 90%|██████████████████████████████▌   | 108/120 [16:37<01:55,  9.66s/it]
 91%|██████████████████████████████▉   | 109/120 [16:45<01:41,  9.25s/it]
 92%|███████████████████████████████▏  | 110/120 [16:53<01:29,  8.96s/it]
                                                                         
{'loss': 4.977, 'grad_norm': 1.2357864379882812, 'learning_rate': 4.894348370484648e-07, 'ppl': 145.03861, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 236.01318359375, 'tokens/total': 1769472, 'tokens/trainable': 800593, 'epoch': 8.47}
-
 92%|███████████████████████████████▏  | 110/120 [16:53<01:29,  8.96s/it]
 92%|███████████████████████████████▍  | 111/120 [17:02<01:18,  8.75s/it]
 93%|███████████████████████████████▋  | 112/120 [17:10<01:08,  8.60s/it]
 94%|████████████████████████████████  | 113/120 [17:18<00:59,  8.50s/it]
 95%|████████████████████████████████▎ | 114/120 [17:26<00:50,  8.43s/it]
 96%|████████████████████████████████▌ | 115/120 [17:35<00:41,  8.38s/it]
                                                                         
{'loss': 4.9484, 'grad_norm': 1.248633623123169, 'learning_rate': 1.464616415230702e-07, 'ppl': 140.94926, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 237.99766540527344, 'tokens/total': 1851392, 'tokens/trainable': 838072, 'epoch': 8.86}
-
 96%|████████████████████████████████▌ | 115/120 [17:35<00:41,  8.38s/it]
 97%|████████████████████████████████▊ | 116/120 [17:43<00:33,  8.34s/it]
 98%|█████████████████████████████████▏| 117/120 [17:49<00:23,  7.72s/it]
 98%|█████████████████████████████████▍| 118/120 [17:59<00:16,  8.47s/it]
 99%|█████████████████████████████████▋| 119/120 [18:08<00:08,  8.41s/it]
100%|██████████████████████████████████| 120/120 [18:16<00:00,  8.37s/it]
                                                                         
{'loss': 4.9698, 'grad_norm': 1.1915208101272583, 'learning_rate': 4.0780717181077015e-09, 'ppl': 143.99808, 'memory/max_active (GiB)': 30.06, 'memory/max_allocated (GiB)': 30.06, 'memory/device_reserved (GiB)': 31.84, 'tokens/train_per_sec_per_gpu': 231.05690002441406, 'tokens/total': 1929216, 'tokens/trainable': 872330, 'epoch': 9.24}
-
100%|██████████████████████████████████| 120/120 [18:16<00:00,  8.37s/it][2026-01-24 13:11:10,448] [INFO] [axolotl.core.trainers.base._save:721] [PID:6937] Saving model checkpoint to ./phi4_african_history_lora_out/checkpoint-120
-
                                                                         
{'train_runtime': 1100.0978, 'train_samples_per_second': 0.873, 'train_steps_per_second': 0.109, 'train_loss': 5.385554504394531, 'memory/max_active (GiB)': 7.18, 'memory/max_allocated (GiB)': 7.18, 'memory/device_reserved (GiB)': 31.84, 'epoch': 9.24, 'tokens/train_per_sec_per_gpu': 0.0}
-
100%|██████████████████████████████████| 120/120 [18:17<00:00,  8.37s/it]
100%|██████████████████████████████████| 120/120 [18:17<00:00,  9.14s/it]
-[2026-01-24 13:11:17,545] [INFO] [axolotl.train.save_trained_model:233] [PID:6937] Training completed! Saving trained model to ./phi4_african_history_lora_out.
-[2026-01-24 13:11:17,908] [INFO] [axolotl.train.save_trained_model:351] [PID:6937] Model successfully saved to ./phi4_african_history_lora_out
-[2026-01-24 13:11:18,131] [INFO] [axolotl.core.trainers.base._save:721] [PID:6937] Saving model checkpoint to ./phi4_african_history_lora_out
-
Processing Files (0 / 0)      : |           |  0.00B /  0.00B            
-
New Data Upload               : |           |  0.00B /  0.00B            
-
-
  ...ora_out/training_args.bin: 100%|███████| 7.76kB / 7.76kB            
-
-
-
  ...adapter_model.safetensors: 100%|███████| 6.30MB / 6.30MB            
-
-
-
-
  ...y_lora_out/tokenizer.json: 100%|███████| 15.5MB / 15.5MB            
-
-
  ...ora_out/training_args.bin: 100%|███████| 7.76kB / 7.76kB            
-
-
-
  ...adapter_model.safetensors: 100%|███████| 6.30MB / 6.30MB            
-
-
-
-
  ...y_lora_out/tokenizer.json: 100%|███████| 15.5MB / 15.5MB            
Processing Files (3 / 3)      : 100%|███████| 21.8MB / 21.8MB,   ???B/s  
-
-
  ...ora_out/training_args.bin: 100%|███████| 7.76kB / 7.76kB            
-
-
-
  ...adapter_model.safetensors: 100%|███████| 6.30MB / 6.30MB            
-
-
-
-
  ...y_lora_out/tokenizer.json: 100%|███████| 15.5MB / 15.5MB            
-
-
  ...ora_out/training_args.bin: 100%|███████| 7.76kB / 7.76kB            
-
-
-
  ...adapter_model.safetensors: 100%|███████| 6.30MB / 6.30MB            
-
-
-
-
  ...y_lora_out/tokenizer.json: 100%|███████| 15.5MB / 15.5MB            
Processing Files (3 / 3)      : 100%|███████| 21.8MB / 21.8MB,  0.00B/s  
-
New Data Upload               : |           |  0.00B /  0.00B,  0.00B/s  
-
  ...ora_out/training_args.bin: 100%|███████| 7.76kB / 7.76kB            
-
  ...adapter_model.safetensors: 100%|███████| 6.30MB / 6.30MB            
-
  ...y_lora_out/tokenizer.json: 100%|███████| 15.5MB / 15.5MB            
\ No newline at end of file
+
                                                                         [2026-01-24 13:42:01,206] [INFO] [axolotl.core.trainers.base._save:721] [PID:9359] Saving model checkpoint to ./phi4_african_history_lora_out/checkpoint-100