Instructions to use c4tdr0ut/Helios-7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use c4tdr0ut/Helios-7B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="c4tdr0ut/Helios-7B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("c4tdr0ut/Helios-7B")
model = AutoModelForCausalLM.from_pretrained("c4tdr0ut/Helios-7B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Grok

How to use c4tdr0ut/Helios-7B with Grok:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use c4tdr0ut/Helios-7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "c4tdr0ut/Helios-7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "c4tdr0ut/Helios-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/c4tdr0ut/Helios-7B

SGLang

How to use c4tdr0ut/Helios-7B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "c4tdr0ut/Helios-7B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "c4tdr0ut/Helios-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "c4tdr0ut/Helios-7B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "c4tdr0ut/Helios-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use c4tdr0ut/Helios-7B with Docker Model Runner:
```
docker model run hf.co/c4tdr0ut/Helios-7B
```

Helios-7B / debug.log

c4tdr0ut

Training in progress, step 24

8186f25 verified about 2 months ago

Raw History Blame Contribute Delete

64.5 kB

	[2026-08-11 19:02:12,745] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:31] baseline 0.000GB ()
	[2026-08-11 19:02:12,747] [INFO] [axolotl.cli.config.load_cfg:336] [PID:31] config:
	{
	"activation_offloading": false,
	"attn_decontaminates_packing": true,
	"attn_implementation": "flash_attention_2",
	"attn_needs_dtype_cast": true,
	"attn_supports_packing": true,
	"attn_uses_flash_lib": true,
	"auto_resume_from_checkpoints": false,
	"axolotl_config_path": "/workspace/task_08e727bf-a87d-47c2-97d0-de595c57c5dc_output/sft_run/sft_training_config.yaml",
	"base_model": "c4tdr0ut/heli-pretrain",
	"base_model_config": "c4tdr0ut/heli-pretrain",
	"batch_size": 150,
	"bf16": true,
	"capabilities": {
	"bf16": true,
	"compute_capability": "sm_90",
	"fp8": true,
	"n_gpu": 1,
	"n_node": 1,
	"tf32": true
	},
	"chat_template": "chatml",
	"context_parallel_size": 1,
	"dataloader_num_workers": 1,
	"dataloader_pin_memory": true,
	"dataloader_prefetch_factor": 256,
	"dataset_num_proc": 17,
	"dataset_prepared_path": "last_finetune_prepared",
	"datasets": [
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "axolotl_rag_conversations_inputs.jsonl",
	"trust_remote_code": false,
	"type": "input_output"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "axolotl_correction_conversations_inputs.json",
	"trust_remote_code": false,
	"type": "input_output"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "pretraining_subset_1515812.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "factual_sft_completion/combined_all_0.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "factual_sft_completion/combined_all_2.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "factual_sft_completion/combined_all_1.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "generic_sft_completion/Augmentoolkit-Augmentoolkit-Pippa-Thoughts_748413.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "generic_sft_completion/Augmentoolkit-Augmentoolkit-Bluemoon-1mil-thoughts_748413.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "generic_sft_completion/Augmentoolkit-Openthoughts-100mil-DifferentFormat_2993654.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "generic_sft_completion/Augmentoolkit-Augmentoolkit-LMsys-800k-Thoughts_748413.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "generic_sft_completion/Augmentoolkit-Augmentoolkit-Generic-Grabbag-Thoughts_1496827.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	},
	{
	"message_property_mappings": {
	"content": "content",
	"role": "role"
	},
	"path": "generic_sft_completion/Augmentoolkit-Augmentoolkit-Capybara-2point5mil-Thoughts_748413.jsonl",
	"trust_remote_code": false,
	"type": "completion"
	}
	],
	"ddp": false,
	"device": "cuda:0",
	"device_map": "auto",
	"dion_rank_fraction": 1.0,
	"dion_rank_multiple_of": 1,
	"eaft_alpha": 1.0,
	"eaft_k": 20,
	"env_capabilities": {
	"torch_version": "2.12.0"
	},
	"eval_batch_size": 4,
	"eval_causal_lm_metrics": [
	"sacrebleu",
	"comet",
	"ter",
	"chrf"
	],
	"eval_max_new_tokens": 128,
	"eval_sample_packing": false,
	"eval_steps": 0.25,
	"eval_table_size": 0,
	"evals_per_epoch": 1,
	"experimental_skip_move_to_device": true,
	"fp16": false,
	"generate_samples": false,
	"generation_do_sample": true,
	"generation_max_new_tokens": 50,
	"generation_prompt_ratio": 0.5,
	"generation_temperature": 0.7,
	"gradient_accumulation_steps": 10,
	"gradient_checkpointing": true,
	"gradient_checkpointing_kwargs": {
	"use_reentrant": true
	},
	"group_by_length": false,
	"hub_model_id": "c4tdr0ut/Helios-7B",
	"hub_strategy": "all_checkpoints",
	"include_tkps": true,
	"is_falcon_derived_model": false,
	"is_llama_derived_model": false,
	"is_mistral_derived_model": true,
	"layer_offloading": false,
	"learning_rate": 2e-05,
	"liger_fused_linear_cross_entropy": true,
	"liger_glu_activation": true,
	"liger_layer_norm": true,
	"liger_rms_norm": true,
	"liger_rope": true,
	"lisa_layers_attribute": "model.layers",
	"load_best_model_at_end": false,
	"load_in_4bit": false,
	"load_in_8bit": false,
	"local_rank": 0,
	"logging_steps": 1,
	"lora_dropout": 0.0,
	"loraplus_lr_embedding": 1e-06,
	"lr_scheduler": "constant",
	"mean_resizing_embeddings": false,
	"merge_method": "memory_efficient",
	"micro_batch_size": 15,
	"model_config_type": "mistral",
	"neftune_noise_alpha": 5.0,
	"num_epochs": 4.0,
	"num_generation_samples": 3,
	"optimizer": "paged_adamw_8bit",
	"otel_metrics_host": "localhost",
	"otel_metrics_port": 8000,
	"output_dir": "./finetune-model-output",
	"pad_to_sequence_len": false,
	"plugins": [
	"axolotl.integrations.liger.LigerPlugin"
	],
	"pretrain_multipack_attn": true,
	"profiler_steps_start": 0,
	"qgalore_cos_threshold": 0.4,
	"qgalore_gamma_proj": 2,
	"qgalore_proj_bits": 4,
	"qgalore_proj_group_size": 256,
	"qgalore_proj_quant": true,
	"qgalore_proj_type": "std",
	"qgalore_queue_size": 5,
	"qgalore_rank": 256,
	"qgalore_scale": 0.25,
	"qgalore_update_proj_gap": 200,
	"qlora_sharded_model_loading": false,
	"quantize_moe_experts": false,
	"ray_num_workers": 1,
	"relora_prune_method": "magnitude",
	"remove_unused_columns": false,
	"resources_per_worker": {
	"GPU": 1
	},
	"sample_packing": true,
	"sample_packing_bin_size": 200,
	"sample_packing_group_size": 100000,
	"save_only_model": false,
	"save_safetensors": true,
	"save_steps": 0.25,
	"save_total_limit": 2,
	"saves_per_epoch": 1,
	"seed": 1337,
	"sequence_len": 5000,
	"shuffle_before_merging_datasets": false,
	"shuffle_merged_datasets": true,
	"skip_prepare_dataset": false,
	"special_tokens": {
	"pad_token": "<unk>"
	},
	"streaming_multipack_buffer_size": 10000,
	"strict": false,
	"tensor_parallel_size": 1,
	"tf32": false,
	"tiled_mlp_use_original_mlp": true,
	"tokenizer_config": "c4tdr0ut/heli-pretrain",
	"tokenizer_save_jinja_files": true,
	"tokenizer_type": "AutoTokenizer",
	"torch_dtype": "torch.bfloat16",
	"train_on_inputs": false,
	"trl": {
	"async_prefetch": false,
	"log_completions": false,
	"mask_truncated_completions": false,
	"ref_model_mixup_alpha": 0.9,
	"ref_model_sync_steps": 64,
	"replay_buffer_size": 0,
	"replay_recompute_logps": true,
	"reroll_max_groups": 1,
	"reroll_start_fraction": 1.0,
	"reward_num_workers": 1,
	"scale_rewards": true,
	"skip_zero_advantage_batches": true,
	"sync_ref_model": false,
	"use_data_producer": false,
	"use_vllm": false,
	"vllm_lora_sync": false,
	"vllm_server_host": "0.0.0.0",
	"vllm_server_port": 8000
	},
	"type_of_model": "AutoModelForCausalLM",
	"use_otel_metrics": false,
	"use_ray": false,
	"val_set_size": 0.04,
	"vllm": {
	"device": "auto",
	"dtype": "auto",
	"gpu_memory_utilization": 0.9,
	"host": "0.0.0.0",
	"port": 8000
	},
	"warmup_ratio": 0.1,
	"weight_decay": 0.0,
	"world_size": 1
	}
	[2026-08-11 19:02:17,597] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:311] [PID:31] EOS: 2 / </s>
	[2026-08-11 19:02:17,597] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:312] [PID:31] BOS: 1 / <s>
	[2026-08-11 19:02:17,598] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:313] [PID:31] PAD: 0 / <unk>
	[2026-08-11 19:02:17,598] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:314] [PID:31] UNK: 0 / <unk>
	[2026-08-11 19:02:17,601] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:543] [PID:31] Unable to find prepared dataset in last_finetune_prepared/5eab9bc891e347cf10de8154a14a218d
	[2026-08-11 19:02:17,601] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:31] Loading raw datasets...
	[2026-08-11 19:02:17,601] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:31] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`.
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 500 examples [00:00, 9589.65 examples/s]
	[2026-08-11 19:02:18,066] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: axolotl_rag_conversations_inputs.jsonl with base_type: input_output and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/500 [00:00<?, ? examples/s][2026-08-11 19:02:18,496] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /workspace/axolotl-venv/lib/python3.12/site-packages/multiprocess/popen_fork.py:66: DeprecationWarning: This process (pid=31) is multi-threaded, use of fork() may lead to deadlocks in the child.
	self.pid = os.fork()

	Tokenizing Prompts (num_proc=17): 0%\| \| 2/500 [00:01<07:47, 1.06 examples/s] Tokenizing Prompts (num_proc=17): 4%\|▎ \| 18/500 [00:02<00:40, 11.79 examples/s] Tokenizing Prompts (num_proc=17): 7%\|▋ \| 37/500 [00:02<00:16, 27.42 examples/s] Tokenizing Prompts (num_proc=17): 14%\|█▍ \| 72/500 [00:02<00:06, 61.15 examples/s] Tokenizing Prompts (num_proc=17): 23%\|██▎ \| 113/500 [00:02<00:03, 107.20 examples/s] Tokenizing Prompts (num_proc=17): 28%\|██▊ \| 142/500 [00:02<00:02, 133.58 examples/s] Tokenizing Prompts (num_proc=17): 40%\|████ \| 200/500 [00:02<00:01, 213.81 examples/s] Tokenizing Prompts (num_proc=17): 48%\|████▊ \| 238/500 [00:02<00:01, 229.44 examples/s] Tokenizing Prompts (num_proc=17): 58%\|█████▊ \| 288/500 [00:02<00:00, 284.63 examples/s] Tokenizing Prompts (num_proc=17): 65%\|██████▌ \| 327/500 [00:02<00:00, 294.54 examples/s] Tokenizing Prompts (num_proc=17): 73%\|███████▎ \| 365/500 [00:03<00:00, 291.66 examples/s] Tokenizing Prompts (num_proc=17): 82%\|████████▏ \| 409/500 [00:03<00:00, 325.36 examples/s] Tokenizing Prompts (num_proc=17): 90%\|█████████ \| 451/500 [00:03<00:00, 338.40 examples/s] Tokenizing Prompts (num_proc=17): 98%\|█████████▊\| 490/500 [00:03<00:00, 289.79 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 500/500 [00:03<00:00, 129.54 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 388 examples [00:00, 2954.03 examples/s] Generating train split: 388 examples [00:00, 2891.60 examples/s]
	[2026-08-11 19:02:22,531] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: axolotl_correction_conversations_inputs.json with base_type: input_output and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/388 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 4%\|▍ \| 16/388 [00:01<00:43, 8.52 examples/s] Tokenizing Prompts (num_proc=17): 14%\|█▍ \| 54/388 [00:02<00:10, 33.33 examples/s] Tokenizing Prompts (num_proc=17): 27%\|██▋ \| 104/388 [00:02<00:03, 71.73 examples/s] Tokenizing Prompts (num_proc=17): 36%\|███▌ \| 138/388 [00:02<00:02, 88.56 examples/s] Tokenizing Prompts (num_proc=17): 51%\|█████▏ \| 199/388 [00:02<00:01, 147.71 examples/s] Tokenizing Prompts (num_proc=17): 65%\|██████▌ \| 253/388 [00:02<00:00, 196.74 examples/s] Tokenizing Prompts (num_proc=17): 75%\|███████▌ \| 292/388 [00:02<00:00, 208.67 examples/s] Tokenizing Prompts (num_proc=17): 87%\|████████▋ \| 338/388 [00:02<00:00, 251.52 examples/s] Tokenizing Prompts (num_proc=17): 99%\|█████████▊\| 383/388 [00:03<00:00, 287.97 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 388/388 [00:03<00:00, 120.92 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 11 examples [00:00, 602.06 examples/s]
	[2026-08-11 19:02:26,503] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: pretraining_subset_1515812.jsonl with base_type: completion and prompt_style: None
	[2026-08-11 19:02:26,507] [WARNING] [datasets.arrow_dataset.map:3413] [PID:31] num_proc must be <= 11. Reducing num_proc to 11 for dataset of size 11.
	Tokenizing Prompts (num_proc=11): 0%\| \| 0/11 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=11): 9%\|▉ \| 1/11 [00:01<00:15, 1.60s/ examples] Tokenizing Prompts (num_proc=11): 27%\|██▋ \| 3/11 [00:01<00:03, 2.22 examples/s] Tokenizing Prompts (num_proc=11): 45%\|████▌ \| 5/11 [00:01<00:01, 3.83 examples/s] Tokenizing Prompts (num_proc=11): 64%\|██████▎ \| 7/11 [00:02<00:00, 5.18 examples/s] Tokenizing Prompts (num_proc=11): 91%\|█████████ \| 10/11 [00:02<00:00, 7.41 examples/s] Tokenizing Prompts (num_proc=11): 100%\|██████████\| 11/11 [00:02<00:00, 3.69 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 1057 examples [00:00, 34419.58 examples/s]
	[2026-08-11 19:02:30,018] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: factual_sft_completion/combined_all_0.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/1057 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 63/1057 [00:01<00:25, 38.67 examples/s] Tokenizing Prompts (num_proc=17): 12%\|█▏ \| 126/1057 [00:01<00:10, 85.83 examples/s] Tokenizing Prompts (num_proc=17): 30%\|██▉ \| 313/1057 [00:01<00:02, 266.76 examples/s] Tokenizing Prompts (num_proc=17): 41%\|████▏ \| 437/1057 [00:01<00:01, 386.78 examples/s] Tokenizing Prompts (num_proc=17): 59%\|█████▉ \| 623/1057 [00:02<00:00, 551.39 examples/s] Tokenizing Prompts (num_proc=17): 71%\|███████ \| 747/1057 [00:02<00:00, 590.90 examples/s] Tokenizing Prompts (num_proc=17): 94%\|█████████▍\| 995/1057 [00:02<00:00, 793.97 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 1057/1057 [00:02<00:00, 402.54 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 1041 examples [00:00, 32791.87 examples/s]
	[2026-08-11 19:02:33,121] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: factual_sft_completion/combined_all_2.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/1041 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 62/1041 [00:01<00:27, 36.23 examples/s] Tokenizing Prompts (num_proc=17): 18%\|█▊ \| 186/1041 [00:01<00:06, 125.98 examples/s] Tokenizing Prompts (num_proc=17): 36%\|███▌ \| 370/1041 [00:02<00:02, 275.52 examples/s] Tokenizing Prompts (num_proc=17): 53%\|█████▎ \| 553/1041 [00:02<00:01, 432.87 examples/s] Tokenizing Prompts (num_proc=17): 71%\|███████ \| 736/1041 [00:02<00:00, 562.13 examples/s] Tokenizing Prompts (num_proc=17): 88%\|████████▊ \| 919/1041 [00:02<00:00, 706.13 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 1041/1041 [00:02<00:00, 393.82 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 1041 examples [00:00, 34785.73 examples/s]
	[2026-08-11 19:02:36,287] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: factual_sft_completion/combined_all_1.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/1041 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 62/1041 [00:01<00:25, 38.17 examples/s] Tokenizing Prompts (num_proc=17): 24%\|██▍ \| 248/1041 [00:01<00:04, 173.56 examples/s] Tokenizing Prompts (num_proc=17): 36%\|███▌ \| 370/1041 [00:01<00:02, 269.79 examples/s] Tokenizing Prompts (num_proc=17): 53%\|█████▎ \| 553/1041 [00:02<00:01, 427.79 examples/s] Tokenizing Prompts (num_proc=17): 71%\|███████ \| 736/1041 [00:02<00:00, 556.70 examples/s] Tokenizing Prompts (num_proc=17): 94%\|█████████▍\| 980/1041 [00:02<00:00, 796.34 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 1041/1041 [00:02<00:00, 402.39 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 40 examples [00:00, 4102.51 examples/s]
	[2026-08-11 19:02:39,388] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: generic_sft_completion/Augmentoolkit-Augmentoolkit-Pippa-Thoughts_748413.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/40 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 8%\|▊ \| 3/40 [00:01<00:16, 2.22 examples/s] Tokenizing Prompts (num_proc=17): 22%\|██▎ \| 9/40 [00:01<00:04, 6.99 examples/s] Tokenizing Prompts (num_proc=17): 35%\|███▌ \| 14/40 [00:01<00:02, 9.88 examples/s] Tokenizing Prompts (num_proc=17): 52%\|█████▎ \| 21/40 [00:01<00:01, 16.52 examples/s] Tokenizing Prompts (num_proc=17): 62%\|██████▎ \| 25/40 [00:02<00:00, 19.74 examples/s] Tokenizing Prompts (num_proc=17): 85%\|████████▌ \| 34/40 [00:02<00:00, 30.94 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 40/40 [00:02<00:00, 16.80 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 45 examples [00:00, 3857.74 examples/s]
	[2026-08-11 19:02:42,226] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: generic_sft_completion/Augmentoolkit-Augmentoolkit-Bluemoon-1mil-thoughts_748413.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/45 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 7%\|▋ \| 3/45 [00:01<00:21, 1.91 examples/s] Tokenizing Prompts (num_proc=17): 13%\|█▎ \| 6/45 [00:01<00:09, 4.22 examples/s] Tokenizing Prompts (num_proc=17): 40%\|████ \| 18/45 [00:01<00:01, 15.84 examples/s] Tokenizing Prompts (num_proc=17): 60%\|██████ \| 27/45 [00:01<00:00, 23.37 examples/s] Tokenizing Prompts (num_proc=17): 78%\|███████▊ \| 35/45 [00:02<00:00, 27.52 examples/s] Tokenizing Prompts (num_proc=17): 96%\|█████████▌\| 43/45 [00:02<00:00, 28.48 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 45/45 [00:02<00:00, 17.38 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 389 examples [00:00, 15786.22 examples/s]
	[2026-08-11 19:02:45,300] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: generic_sft_completion/Augmentoolkit-Openthoughts-100mil-DifferentFormat_2993654.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/389 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 23/389 [00:01<00:27, 13.46 examples/s] Tokenizing Prompts (num_proc=17): 18%\|█▊ \| 69/389 [00:01<00:06, 46.97 examples/s] Tokenizing Prompts (num_proc=17): 24%\|██▎ \| 92/389 [00:01<00:04, 64.34 examples/s] Tokenizing Prompts (num_proc=17): 41%\|████▏ \| 161/389 [00:02<00:01, 133.17 examples/s] Tokenizing Prompts (num_proc=17): 53%\|█████▎ \| 207/389 [00:02<00:01, 179.30 examples/s] Tokenizing Prompts (num_proc=17): 65%\|██████▌ \| 253/389 [00:02<00:00, 218.99 examples/s] Tokenizing Prompts (num_proc=17): 83%\|████████▎ \| 322/389 [00:02<00:00, 294.81 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 389/389 [00:02<00:00, 146.34 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 813 examples [00:00, 70109.16 examples/s]
	[2026-08-11 19:02:48,430] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: generic_sft_completion/Augmentoolkit-Augmentoolkit-LMsys-800k-Thoughts_748413.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/813 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 48/813 [00:01<00:24, 31.02 examples/s] Tokenizing Prompts (num_proc=17): 18%\|█▊ \| 144/813 [00:01<00:06, 108.15 examples/s] Tokenizing Prompts (num_proc=17): 30%\|██▉ \| 240/813 [00:01<00:02, 196.60 examples/s] Tokenizing Prompts (num_proc=17): 41%\|████▏ \| 336/813 [00:01<00:01, 289.09 examples/s] Tokenizing Prompts (num_proc=17): 53%\|█████▎ \| 432/813 [00:02<00:01, 360.93 examples/s] Tokenizing Prompts (num_proc=17): 71%\|███████ \| 576/813 [00:02<00:00, 521.34 examples/s] Tokenizing Prompts (num_proc=17): 83%\|████████▎ \| 672/813 [00:02<00:00, 603.26 examples/s] Tokenizing Prompts (num_proc=17): 94%\|█████████▍\| 766/813 [00:02<00:00, 660.00 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 813/813 [00:02<00:00, 323.82 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 2531 examples [00:00, 97320.19 examples/s]
	[2026-08-11 19:02:51,457] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: generic_sft_completion/Augmentoolkit-Augmentoolkit-Generic-Grabbag-Thoughts_1496827.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/2531 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 149/2531 [00:02<00:33, 70.55 examples/s] Tokenizing Prompts (num_proc=17): 41%\|████ \| 1043/2531 [00:02<00:02, 509.55 examples/s] Tokenizing Prompts (num_proc=17): 53%\|█████▎ \| 1341/2531 [00:02<00:01, 670.58 examples/s] Tokenizing Prompts (num_proc=17): 65%\|██████▍ \| 1639/2531 [00:02<00:01, 847.87 examples/s] Tokenizing Prompts (num_proc=17): 82%\|████████▏ \| 2086/2531 [00:02<00:00, 1192.76 examples/s] Tokenizing Prompts (num_proc=17): 94%\|█████████▍\| 2383/2531 [00:03<00:00, 1414.16 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 2531/2531 [00:03<00:00, 787.24 examples/s]
	Generating train split: 0 examples [00:00, ? examples/s] Generating train split: 444 examples [00:00, 41194.31 examples/s]
	[2026-08-11 19:02:55,122] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:31] Loading dataset: generic_sft_completion/Augmentoolkit-Augmentoolkit-Capybara-2point5mil-Thoughts_748413.jsonl with base_type: completion and prompt_style: None
	Tokenizing Prompts (num_proc=17): 0%\| \| 0/444 [00:00<?, ? examples/s] Tokenizing Prompts (num_proc=17): 6%\|▌ \| 27/444 [00:01<00:24, 17.18 examples/s] Tokenizing Prompts (num_proc=17): 18%\|█▊ \| 80/444 [00:01<00:06, 58.21 examples/s] Tokenizing Prompts (num_proc=17): 36%\|███▌ \| 158/444 [00:01<00:02, 124.21 examples/s] Tokenizing Prompts (num_proc=17): 47%\|████▋ \| 210/444 [00:01<00:01, 167.76 examples/s] Tokenizing Prompts (num_proc=17): 59%\|█████▉ \| 262/444 [00:02<00:00, 216.33 examples/s] Tokenizing Prompts (num_proc=17): 71%\|███████ \| 314/444 [00:02<00:00, 264.54 examples/s] Tokenizing Prompts (num_proc=17): 82%\|████████▏ \| 366/444 [00:02<00:00, 313.18 examples/s] Tokenizing Prompts (num_proc=17): 94%\|█████████▍\| 418/444 [00:02<00:00, 346.24 examples/s] Tokenizing Prompts (num_proc=17): 100%\|██████████\| 444/444 [00:02<00:00, 170.98 examples/s]
	[2026-08-11 19:02:57,993] [INFO] [axolotl.utils.data.shared.merge_datasets:712] [PID:31] Merging datasets...
	[2026-08-11 19:02:58,037] [DEBUG] [axolotl.utils.data.shared.merge_datasets:716] [PID:31] Shuffling merged datasets...
	[2026-08-11 19:02:58,273] [INFO] [axolotl.utils.data.utils._log_dataset_stats:212] [PID:31] min_input_len: 8
	[2026-08-11 19:02:58,274] [INFO] [axolotl.utils.data.utils._log_dataset_stats:213] [PID:31] max_input_len: 15017
	Dropping Invalid Sequences (<None or >5000) (num_proc=17): 0%\| \| 0/9474 [00:00<?, ? examples/s] Dropping Invalid Sequences (<None or >5000) (num_proc=17): 6%\|▌ \| 558/9474 [00:01<00:30, 294.29 examples/s] Dropping Invalid Sequences (<None or >5000) (num_proc=17): 47%\|████▋ \| 4461/9474 [00:01<00:01, 3002.66 examples/s] Dropping Invalid Sequences (<None or >5000) (num_proc=17): 100%\|██████████\| 9474/9474 [00:02<00:00, 4284.64 examples/s]
	[2026-08-11 19:03:00,622] [INFO] [axolotl.utils.data.utils._drop_outside_range:306] [PID:31] Dropped 443 sequences outside valid range ([None, 5000])
	[2026-08-11 19:03:00,626] [INFO] [axolotl.utils.trainer.process_datasets_for_packing:254] [PID:31] dropping token_type_ids column if it exists
	Drop Samples with Zero Trainable Tokens (num_proc=17): 0%\| \| 0/9031 [00:00<?, ? examples/s] Drop Samples with Zero Trainable Tokens (num_proc=17): 6%\|▌ \| 532/9031 [00:01<00:27, 305.55 examples/s] Drop Samples with Zero Trainable Tokens (num_proc=17): 24%\|██▎ \| 2127/9031 [00:01<00:04, 1471.87 examples/s] Drop Samples with Zero Trainable Tokens (num_proc=17): 88%\|████████▊ \| 7969/9031 [00:01<00:00, 6898.04 examples/s] Drop Samples with Zero Trainable Tokens (num_proc=17): 100%\|██████████\| 9031/9031 [00:02<00:00, 4297.46 examples/s]
	Add position_id column (Sample Packing) (num_proc=17): 0%\| \| 0/9031 [00:00<?, ? examples/s][2026-08-11 19:03:02,994] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /workspace/axolotl-venv/lib/python3.12/site-packages/multiprocess/popen_fork.py:66: DeprecationWarning: This process (pid=31) is multi-threaded, use of fork() may lead to deadlocks in the child.
	self.pid = os.fork()

	Add position_id column (Sample Packing) (num_proc=17): 6%\|▌ \| 532/9031 [00:02<00:34, 249.39 examples/s] Add position_id column (Sample Packing) (num_proc=17): 35%\|███▌ \| 3190/9031 [00:02<00:03, 1898.45 examples/s] Add position_id column (Sample Packing) (num_proc=17): 94%\|█████████▍\| 8500/9031 [00:02<00:00, 6099.61 examples/s] Add position_id column (Sample Packing) (num_proc=17): 100%\|██████████\| 9031/9031 [00:02<00:00, 3623.92 examples/s]
	Saving the dataset (0/17 shards): 0%\| \| 0/9031 [00:00<?, ? examples/s] Saving the dataset (0/17 shards): 6%\|▌ \| 531/9031 [00:05<01:28, 96.46 examples/s] Saving the dataset (1/17 shards): 6%\|▌ \| 531/9031 [00:05<01:28, 96.46 examples/s] Saving the dataset (1/17 shards): 12%\|█▏ \| 1062/9031 [00:05<00:35, 222.35 examples/s] Saving the dataset (2/17 shards): 12%\|█▏ \| 1062/9031 [00:05<00:35, 222.35 examples/s] Saving the dataset (3/17 shards): 18%\|█▊ \| 1594/9031 [00:05<00:33, 222.35 examples/s] Saving the dataset (4/17 shards): 24%\|██▎ \| 2126/9031 [00:05<00:31, 222.35 examples/s] Saving the dataset (5/17 shards): 29%\|██▉ \| 2657/9031 [00:05<00:28, 222.35 examples/s] Saving the dataset (6/17 shards): 35%\|███▌ \| 3189/9031 [00:05<00:26, 222.35 examples/s] Saving the dataset (7/17 shards): 47%\|████▋ \| 4251/9031 [00:05<00:21, 222.35 examples/s] Saving the dataset (8/17 shards): 47%\|████▋ \| 4251/9031 [00:05<00:21, 222.35 examples/s] Saving the dataset (9/17 shards): 53%\|█████▎ \| 4782/9031 [00:05<00:19, 222.35 examples/s] Saving the dataset (10/17 shards): 59%\|█████▉ \| 5313/9031 [00:05<00:16, 222.35 examples/s] Saving the dataset (11/17 shards): 65%\|██████▍ \| 5844/9031 [00:05<00:14, 222.35 examples/s] Saving the dataset (12/17 shards): 71%\|███████ \| 6375/9031 [00:05<00:11, 222.35 examples/s] Saving the dataset (13/17 shards): 76%\|███████▋ \| 6906/9031 [00:05<00:09, 222.35 examples/s] Saving the dataset (14/17 shards): 82%\|████████▏ \| 7437/9031 [00:05<00:07, 222.35 examples/s] Saving the dataset (15/17 shards): 88%\|████████▊ \| 7968/9031 [00:05<00:04, 222.35 examples/s] Saving the dataset (15/17 shards): 94%\|█████████▍\| 8500/9031 [00:05<00:00, 2684.59 examples/s] Saving the dataset (16/17 shards): 94%\|█████████▍\| 8500/9031 [00:05<00:00, 2684.59 examples/s] Saving the dataset (17/17 shards): 100%\|██████████\| 9031/9031 [00:05<00:00, 2684.59 examples/s] Saving the dataset (17/17 shards): 100%\|██████████\| 9031/9031 [00:06<00:00, 1500.92 examples/s]
	[2026-08-11 19:03:11,569] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:418] [PID:31] total_num_tokens: 17_945_275
	[2026-08-11 19:03:11,619] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:440] [PID:31] total_supervised_tokens: 17_669_047
	[2026-08-11 19:03:11,772] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /root/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/popen_fork.py:66: DeprecationWarning: This process (pid=31) is multi-threaded, use of fork() may lead to deadlocks in the child.
	self.pid = os.fork()

	[2026-08-11 19:03:14,907] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.3644943237304688
	[2026-08-11 19:03:16,260] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.352339267730713
	[2026-08-11 19:03:17,635] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.3746399879455566
	[2026-08-11 19:03:19,068] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.432831048965454
	[2026-08-11 19:03:19,109] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:443] [PID:31] gather_len_batches: [240]
	[2026-08-11 19:03:19,110] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:497] [PID:31] data_loader_len: 24
	[2026-08-11 19:03:19,110] [INFO] [axolotl.utils.trainer.calc_sample_packing_eff_est:506] [PID:31] sample_packing_eff_est across ranks: [0.9969597222222222]
	[2026-08-11 19:03:19,110] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:518] [PID:31] sample_packing_eff_est: 1.0
	[2026-08-11 19:03:19,110] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:523] [PID:31] total_num_steps: 96
	[2026-08-11 19:03:19,111] [INFO] [axolotl.utils.data.sft._prepare_standard_dataset:121] [PID:31] Maximum number of steps set at 96
	[2026-08-11 19:03:19,121] [DEBUG] [axolotl.train.setup_model_and_tokenizer:70] [PID:31] loading tokenizer... c4tdr0ut/heli-pretrain
	[2026-08-11 19:03:20,183] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:311] [PID:31] EOS: 2 / </s>
	[2026-08-11 19:03:20,183] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:312] [PID:31] BOS: 1 / <s>
	[2026-08-11 19:03:20,184] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:313] [PID:31] PAD: 0 / <unk>
	[2026-08-11 19:03:20,184] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:314] [PID:31] UNK: 0 / <unk>
	[2026-08-11 19:03:20,184] [DEBUG] [axolotl.train.setup_model_and_tokenizer:81] [PID:31] Loading model
	[2026-08-11 19:03:20,301] [DEBUG] [axolotl.monkeypatch.torchao_optim.patch_torchao_optim_state_8bit:75] [PID:31] Patched OptimState8bit for torch.compile compatibility
	[2026-08-11 19:03:20,301] [DEBUG] [axolotl.monkeypatch.torchao_optim.patch_torchao_optim_state_8bit:122] [PID:31] Patched OptimState4bit for torch.compile compatibility
	[2026-08-11 19:03:20,302] [DEBUG] [axolotl.monkeypatch.torchao_optim.patch_torchao_optim_state_8bit:154] [PID:31] Patched OptimStateFp8 for torch.compile compatibility
	[2026-08-11 19:03:20,310] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:94] [PID:31] Patched Trainer.evaluation_loop with nanmean loss calculation
	[2026-08-11 19:03:20,311] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:148] [PID:31] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation
	[2026-08-11 19:03:20,312] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:912] [PID:31] Applying multipack dataloader patch for sample packing...
	[2026-08-11 19:03:20,322] [INFO] [axolotl.monkeypatch.attention.flash_attn_4.fa4_usable:140] [PID:31] Flash Attention 4 is available for your GPU and offers faster training. To enable: pip install --pre flash-attn-4
	[2026-08-11 19:03:24,678] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /workspace/axolotl-venv/lib/python3.12/site-packages/torch/jit/_script.py:1488: DeprecationWarning: `torch.jit.script` is deprecated. Please switch to `torch.compile` or `torch.export`.
	warnings.warn(

	[2026-08-11 19:03:24,784] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /workspace/axolotl-venv/lib/python3.12/site-packages/fla/ops/nsa/parallel.py:20: ImportWarning: Flash Attention is not installed. Please install it via `pip install flash-attn --no-build-isolation`
	warnings.warn(

	[2026-08-11 19:03:25,005] [INFO] [axolotl.integrations.liger.plugin.pre_model_load:145] [PID:31] Applying LIGER to mistral with kwargs: {'rope': True, 'cross_entropy': None, 'fused_linear_cross_entropy': True, 'rms_norm': True, 'swiglu': True}
	[2026-08-11 19:04:14,342] [WARNING] [kernels._versions.resolve_version_spec_as_ref:83] [PID:31] You are using version 1 of 'kernels-community/flash-attn2', but version 3 is available.
	Fetching ... files: 0it [00:00, ?it/s] Fetching ... files: 1it [00:00, 8.89it/s] Fetching ... files: 2it [00:05, 3.05s/it] Fetching ... files: 23it [00:05, 4.40it/s]
	Loading weights: 0%\| \| 0/291 [00:00<?, ?it/s] Loading weights: 1%\| \| 2/291 [00:00<00:17, 16.34it/s] Loading weights: 5%\|▌ \| 15/291 [00:00<00:03, 70.76it/s] Loading weights: 11%\|█ \| 31/291 [00:00<00:02, 100.08it/s] Loading weights: 14%\|█▍ \| 42/291 [00:00<00:02, 98.29it/s] Loading weights: 20%\|█▉ \| 58/291 [00:00<00:02, 115.07it/s] Loading weights: 24%\|██▍ \| 70/291 [00:00<00:01, 111.24it/s] Loading weights: 29%\|██▉ \| 85/291 [00:00<00:01, 116.40it/s] Loading weights: 33%\|███▎ \| 97/291 [00:00<00:01, 108.51it/s] Loading weights: 38%\|███▊ \| 112/291 [00:01<00:01, 116.78it/s] Loading weights: 43%\|████▎ \| 124/291 [00:01<00:01, 112.18it/s] Loading weights: 48%\|████▊ \| 139/291 [00:01<00:01, 117.28it/s] Loading weights: 52%\|█████▏ \| 151/291 [00:01<00:01, 111.38it/s] Loading weights: 57%\|█████▋ \| 166/291 [00:01<00:01, 117.88it/s] Loading weights: 61%\|██████ \| 178/291 [00:01<00:01, 111.62it/s] Loading weights: 65%\|██████▌ \| 190/291 [00:01<00:00, 112.22it/s] Loading weights: 69%\|██████▉ \| 202/291 [00:01<00:00, 111.04it/s] Loading weights: 74%\|███████▎ \| 214/291 [00:01<00:00, 107.60it/s] Loading weights: 79%\|███████▊ \| 229/291 [00:02<00:00, 115.35it/s] Loading weights: 83%\|████████▎ \| 241/291 [00:02<00:00, 110.60it/s] Loading weights: 88%\|████████▊ \| 256/291 [00:02<00:00, 116.51it/s] Loading weights: 92%\|█████████▏\| 268/291 [00:02<00:00, 111.35it/s] Loading weights: 97%\|█████████▋\| 283/291 [00:02<00:00, 116.71it/s] Loading weights: 100%\|██████████\| 291/291 [00:02<00:00, 110.96it/s]
	[2026-08-11 19:04:23,372] [INFO] [axolotl.loaders.model._configure_embedding_dtypes:482] [PID:31] Converting modules to torch.bfloat16
	[2026-08-11 19:04:23,375] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:31] Memory usage after model load 14.251GB (+14.251GB allocated, +14.502GB reserved)
	[2026-08-11 19:04:24,900] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /workspace/axolotl-venv/lib/python3.12/site-packages/torch/jit/_script.py:1488: DeprecationWarning: `torch.jit.script` is deprecated. Please switch to `torch.compile` or `torch.export`.
	warnings.warn(

	[2026-08-11 19:04:25,089] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /workspace/axolotl-venv/lib/python3.12/site-packages/torch/jit/_script.py:1640: DeprecationWarning: `torch.jit.interface` is deprecated. Please use `torch.compile` instead.
	warnings.warn(

	[2026-08-11 19:04:25,356] [WARNING] [accelerate.utils.other.check_os_kernel:520] [PID:31] [RANK 0] Detected kernel version 4.19.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
	[2026-08-11 19:04:27,407] [INFO] [axolotl.train.save_initial_configs:486] [PID:31] Pre-saving tokenizer to ./finetune-model-output...
	[2026-08-11 19:04:27,448] [INFO] [axolotl.train.save_initial_configs:491] [PID:31] Pre-saving model config to ./finetune-model-output...
	[2026-08-11 19:04:27,453] [INFO] [axolotl.train.execute_training:234] [PID:31] Starting trainer...
	[2026-08-11 19:04:28,034] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /root/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/popen_fork.py:66: DeprecationWarning: This process (pid=31) is multi-threaded, use of fork() may lead to deadlocks in the child.
	self.pid = os.fork()

	[2026-08-11 19:04:30,659] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.315068244934082
	[2026-08-11 19:04:31,972] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.3127715587615967
	[2026-08-11 19:04:33,341] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.368100643157959
	[2026-08-11 19:04:34,639] [DEBUG] [axolotl.utils.samplers.multipack.__len__:467] [PID:31] generate_batches time: 1.297865867614746
	[2026-08-11 19:04:34,639] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:443] [PID:31] gather_len_batches: [240]
	0%\| \| 0/96 [00:00<?, ?it/s][2026-08-11 19:04:34,682] [INFO] [axolotl.core.trainers.base.evaluate:464] [PID:31] Running evaluation step...

	0%\| \| 0/90 [00:00<?, ?it/s][A
	2%\|▏ \| 2/90 [00:00<00:28, 3.05it/s][A
	4%\|▍ \| 4/90 [00:01<00:26, 3.30it/s][A
	6%\|▌ \| 5/90 [00:01<00:29, 2.88it/s][A
	7%\|▋ \| 6/90 [00:02<00:33, 2.51it/s][A
	8%\|▊ \| 7/90 [00:02<00:28, 2.95it/s][A
	9%\|▉ \| 8/90 [00:02<00:32, 2.54it/s][A
	10%\|█ \| 9/90 [00:03<00:30, 2.65it/s][A
	11%\|█ \| 10/90 [00:03<00:27, 2.90it/s][A
	12%\|█▏ \| 11/90 [00:03<00:27, 2.83it/s][A
	13%\|█▎ \| 12/90 [00:04<00:30, 2.53it/s][A
	14%\|█▍ \| 13/90 [00:04<00:33, 2.32it/s][A
	16%\|█▌ \| 14/90 [00:05<00:34, 2.20it/s][A
	17%\|█▋ \| 15/90 [00:05<00:35, 2.11it/s][A
	18%\|█▊ \| 16/90 [00:06<00:27, 2.64it/s][A
	19%\|█▉ \| 17/90 [00:06<00:30, 2.39it/s][A
	20%\|██ \| 18/90 [00:07<00:32, 2.22it/s][A
	21%\|██ \| 19/90 [00:07<00:33, 2.14it/s][A
	22%\|██▏ \| 20/90 [00:08<00:33, 2.06it/s][A
	23%\|██▎ \| 21/90 [00:08<00:32, 2.14it/s][A
	26%\|██▌ \| 23/90 [00:09<00:25, 2.64it/s][A
	27%\|██▋ \| 24/90 [00:09<00:27, 2.40it/s][A
	28%\|██▊ \| 25/90 [00:10<00:28, 2.26it/s][A
	29%\|██▉ \| 26/90 [00:10<00:23, 2.78it/s][A
	30%\|███ \| 27/90 [00:10<00:22, 2.80it/s][A
	32%\|███▏ \| 29/90 [00:11<00:18, 3.30it/s][A
	33%\|███▎ \| 30/90 [00:11<00:19, 3.02it/s][A
	34%\|███▍ \| 31/90 [00:12<00:22, 2.59it/s][A
	36%\|███▌ \| 32/90 [00:12<00:22, 2.54it/s][A
	37%\|███▋ \| 33/90 [00:13<00:24, 2.35it/s][A
	38%\|███▊ \| 34/90 [00:13<00:25, 2.20it/s][A
	39%\|███▉ \| 35/90 [00:13<00:22, 2.40it/s][A
	40%\|████ \| 36/90 [00:14<00:23, 2.34it/s][A
	41%\|████ \| 37/90 [00:14<00:19, 2.65it/s][A
	42%\|████▏ \| 38/90 [00:15<00:20, 2.56it/s][A
	43%\|████▎ \| 39/90 [00:15<00:17, 2.93it/s][A
	44%\|████▍ \| 40/90 [00:15<00:16, 3.02it/s][A
	46%\|████▌ \| 41/90 [00:15<00:17, 2.83it/s][A
	47%\|████▋ \| 42/90 [00:16<00:15, 3.19it/s][A
	48%\|████▊ \| 43/90 [00:16<00:16, 2.85it/s][A
	49%\|████▉ \| 44/90 [00:16<00:14, 3.13it/s][A
	50%\|█████ \| 45/90 [00:17<00:14, 3.04it/s][A
	51%\|█████ \| 46/90 [00:17<00:11, 3.72it/s][A
	52%\|█████▏ \| 47/90 [00:17<00:11, 3.79it/s][A
	53%\|█████▎ \| 48/90 [00:17<00:12, 3.43it/s][A
	54%\|█████▍ \| 49/90 [00:18<00:12, 3.34it/s][A
	56%\|█████▌ \| 50/90 [00:18<00:11, 3.57it/s][A
	57%\|█████▋ \| 51/90 [00:18<00:13, 2.89it/s][A
	58%\|█████▊ \| 52/90 [00:19<00:12, 3.01it/s][A
	59%\|█████▉ \| 53/90 [00:19<00:14, 2.58it/s][A
	60%\|██████ \| 54/90 [00:20<00:12, 2.88it/s][A
	61%\|██████ \| 55/90 [00:20<00:13, 2.51it/s][A
	62%\|██████▏ \| 56/90 [00:20<00:12, 2.62it/s][A
	63%\|██████▎ \| 57/90 [00:21<00:10, 3.28it/s][A
	64%\|██████▍ \| 58/90 [00:21<00:08, 3.80it/s][A
	66%\|██████▌ \| 59/90 [00:21<00:10, 2.95it/s][A
	67%\|██████▋ \| 60/90 [00:22<00:11, 2.54it/s][A
	68%\|██████▊ \| 61/90 [00:22<00:11, 2.60it/s][A
	69%\|██████▉ \| 62/90 [00:23<00:11, 2.45it/s][A
	70%\|███████ \| 63/90 [00:23<00:09, 2.80it/s][A
	71%\|███████ \| 64/90 [00:23<00:07, 3.39it/s][A
	72%\|███████▏ \| 65/90 [00:23<00:09, 2.75it/s][A
	73%\|███████▎ \| 66/90 [00:24<00:09, 2.47it/s][A
	74%\|███████▍ \| 67/90 [00:24<00:08, 2.61it/s][A
	76%\|███████▌ \| 68/90 [00:25<00:09, 2.37it/s][A
	77%\|███████▋ \| 69/90 [00:25<00:08, 2.61it/s][A
	78%\|███████▊ \| 70/90 [00:25<00:06, 2.94it/s][A
	79%\|███████▉ \| 71/90 [00:26<00:06, 2.75it/s][A
	80%\|████████ \| 72/90 [00:26<00:05, 3.12it/s][A
	81%\|████████ \| 73/90 [00:26<00:05, 3.31it/s][A
	82%\|████████▏ \| 74/90 [00:27<00:05, 2.71it/s][A
	83%\|████████▎ \| 75/90 [00:27<00:06, 2.42it/s][A
	84%\|████████▍ \| 76/90 [00:28<00:05, 2.69it/s][A
	86%\|████████▌ \| 77/90 [00:28<00:04, 3.10it/s][A
	87%\|████████▋ \| 78/90 [00:28<00:04, 2.84it/s][A
	88%\|████████▊ \| 79/90 [00:29<00:04, 2.47it/s][A
	89%\|████████▉ \| 80/90 [00:29<00:03, 2.73it/s][A
	90%\|█████████ \| 81/90 [00:30<00:03, 2.37it/s][A
	91%\|█████████ \| 82/90 [00:30<00:03, 2.27it/s][A
	92%\|█████████▏\| 83/90 [00:31<00:03, 2.15it/s][A
	93%\|█████████▎\| 84/90 [00:31<00:02, 2.07it/s][A
	94%\|█████████▍\| 85/90 [00:32<00:02, 2.04it/s][A
	96%\|█████████▌\| 86/90 [00:32<00:02, 1.99it/s][A
	97%\|█████████▋\| 87/90 [00:32<00:01, 2.61it/s][A
	98%\|█████████▊\| 88/90 [00:33<00:00, 2.35it/s][A
	99%\|█████████▉\| 89/90 [00:33<00:00, 2.63it/s][A
	100%\|██████████\| 90/90 [00:33<00:00, 2.69it/s][A
	[A{'eval_loss': '1.762', 'eval_runtime': '36.99', 'eval_samples_per_second': '9.787', 'eval_steps_per_second': '2.46', 'eval_ppl': '5.824', 'memory/max_active (GiB)': '19.79', 'memory/max_allocated (GiB)': '19.79', 'memory/device_reserved (GiB)': '26.67', 'epoch': 0}
	0%\| \| 0/96 [00:36<?, ?it/s]
	100%\|██████████\| 90/90 [00:34<00:00, 2.69it/s][A
	[A 1%\| \| 1/96 [02:01<3:12:52, 121.82s/it] {'loss': '1.668', 'grad_norm': '30.88', 'learning_rate': '2e-05', 'ppl': '5.3', 'memory/max_active (GiB)': '57.85', 'memory/max_allocated (GiB)': '57.85', 'memory/device_reserved (GiB)': '60.99', 'tokens/trainable': 742568, 'tokens/total': 749902, 'epoch': '0.04167'}
	1%\| \| 1/96 [02:01<3:12:52, 121.82s/it] 2%\|▏ \| 2/96 [03:20<2:31:24, 96.64s/it] {'loss': '1.606', 'grad_norm': '28.88', 'learning_rate': '2e-05', 'ppl': '4.985', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '65.01', 'tokens/train_per_sec_per_gpu': '9394', 'tokens/trainable': 1484702, 'tokens/total': 1499819, 'epoch': '0.08333'}
	2%\|▏ \| 2/96 [03:20<2:31:24, 96.64s/it] 3%\|▎ \| 3/96 [04:38<2:16:08, 87.83s/it] {'loss': '1.242', 'grad_norm': '19.12', 'learning_rate': '2e-05', 'ppl': '3.461', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '65.01', 'tokens/train_per_sec_per_gpu': '9349', 'tokens/trainable': 2207855, 'tokens/total': 2249708, 'epoch': '0.125'}
	3%\|▎ \| 3/96 [04:38<2:16:08, 87.83s/it] 4%\|▍ \| 4/96 [05:56<2:08:53, 84.06s/it] {'loss': '1.165', 'grad_norm': '33.5', 'learning_rate': '2e-05', 'ppl': '3.206', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '65.01', 'tokens/train_per_sec_per_gpu': '9482', 'tokens/trainable': 2950021, 'tokens/total': 2999475, 'epoch': '0.1667'}
	4%\|▍ \| 4/96 [05:56<2:08:53, 84.06s/it] 5%\|▌ \| 5/96 [07:14<2:03:55, 81.71s/it] {'loss': '1.185', 'grad_norm': '17', 'learning_rate': '2e-05', 'ppl': '3.272', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '65.01', 'tokens/train_per_sec_per_gpu': '9495', 'tokens/trainable': 3686352, 'tokens/total': 3749133, 'epoch': '0.2083'}
	5%\|▌ \| 5/96 [07:14<2:03:55, 81.71s/it] 6%\|▋ \| 6/96 [08:31<2:00:32, 80.36s/it] {'loss': '1.06', 'grad_norm': '8.812', 'learning_rate': '2e-05', 'ppl': '2.885', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '65.01', 'tokens/train_per_sec_per_gpu': '9440', 'tokens/trainable': 4420112, 'tokens/total': 4498747, 'epoch': '0.25'}
	6%\|▋ \| 6/96 [08:31<2:00:32, 80.36s/it] 7%\|▋ \| 7/96 [09:49<1:57:59, 79.55s/it] {'loss': '0.9709', 'grad_norm': '6.844', 'learning_rate': '2e-05', 'ppl': '2.64', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '65.01', 'tokens/train_per_sec_per_gpu': '9429', 'tokens/trainable': 5154407, 'tokens/total': 5248232, 'epoch': '0.2917'}
	7%\|▋ \| 7/96 [09:49<1:57:59, 79.55s/it] 8%\|▊ \| 8/96 [11:07<1:56:02, 79.12s/it] {'loss': '0.9857', 'grad_norm': '4.281', 'learning_rate': '2e-05', 'ppl': '2.68', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9504', 'tokens/trainable': 5897628, 'tokens/total': 5997795, 'epoch': '0.3333'}
	8%\|▊ \| 8/96 [11:07<1:56:02, 79.12s/it] 9%\|▉ \| 9/96 [12:25<1:54:02, 78.65s/it] {'loss': '0.9362', 'grad_norm': '3.953', 'learning_rate': '2e-05', 'ppl': '2.55', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9489', 'tokens/trainable': 6634055, 'tokens/total': 6747342, 'epoch': '0.375'}
	9%\|▉ \| 9/96 [12:25<1:54:02, 78.65s/it] 10%\|█ \| 10/96 [13:43<1:52:20, 78.37s/it] {'loss': '0.9151', 'grad_norm': '4.656', 'learning_rate': '2e-05', 'ppl': '2.497', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9548', 'tokens/trainable': 7376513, 'tokens/total': 7497031, 'epoch': '0.4167'}
	10%\|█ \| 10/96 [13:43<1:52:20, 78.37s/it] 11%\|█▏ \| 11/96 [15:01<1:50:50, 78.24s/it] {'loss': '0.9457', 'grad_norm': '5.469', 'learning_rate': '2e-05', 'ppl': '2.575', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9519', 'tokens/trainable': 8118382, 'tokens/total': 8246701, 'epoch': '0.4583'}
	11%\|█▏ \| 11/96 [15:01<1:50:50, 78.24s/it] 12%\|█▎ \| 12/96 [16:18<1:49:20, 78.10s/it] {'loss': '0.8496', 'grad_norm': '2.969', 'learning_rate': '2e-05', 'ppl': '2.339', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9560', 'tokens/trainable': 8862049, 'tokens/total': 8995758, 'epoch': '0.5'}
	12%\|█▎ \| 12/96 [16:18<1:49:20, 78.10s/it] 14%\|█▎ \| 13/96 [17:36<1:47:44, 77.88s/it] {'loss': '0.8735', 'grad_norm': '3.25', 'learning_rate': '2e-05', 'ppl': '2.395', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9489', 'tokens/trainable': 9596186, 'tokens/total': 9745292, 'epoch': '0.5417'}
	14%\|█▎ \| 13/96 [17:36<1:47:44, 77.88s/it] 15%\|█▍ \| 14/96 [18:53<1:46:18, 77.79s/it] {'loss': '0.8467', 'grad_norm': '2.234', 'learning_rate': '2e-05', 'ppl': '2.332', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9480', 'tokens/trainable': 10331619, 'tokens/total': 10494275, 'epoch': '0.5833'}
	15%\|█▍ \| 14/96 [18:53<1:46:18, 77.79s/it] 16%\|█▌ \| 15/96 [20:11<1:44:48, 77.64s/it] {'loss': '0.9049', 'grad_norm': '2.453', 'learning_rate': '2e-05', 'ppl': '2.472', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9656', 'tokens/trainable': 11077979, 'tokens/total': 11243240, 'epoch': '0.625'}
	16%\|█▌ \| 15/96 [20:11<1:44:48, 77.64s/it] 17%\|█▋ \| 16/96 [21:28<1:43:34, 77.68s/it] {'loss': '0.8693', 'grad_norm': '2.672', 'learning_rate': '2e-05', 'ppl': '2.385', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9484', 'tokens/trainable': 11815398, 'tokens/total': 11991989, 'epoch': '0.6667'}
	17%\|█▋ \| 16/96 [21:28<1:43:34, 77.68s/it] 18%\|█▊ \| 17/96 [22:46<1:42:22, 77.76s/it] {'loss': '0.8343', 'grad_norm': '2.891', 'learning_rate': '2e-05', 'ppl': '2.303', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9529', 'tokens/trainable': 12558179, 'tokens/total': 12741571, 'epoch': '0.7083'}
	18%\|█▊ \| 17/96 [22:46<1:42:22, 77.76s/it] 19%\|█▉ \| 18/96 [24:04<1:40:54, 77.62s/it] {'loss': '0.8577', 'grad_norm': '2.438', 'learning_rate': '2e-05', 'ppl': '2.358', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9480', 'tokens/trainable': 13291084, 'tokens/total': 13490669, 'epoch': '0.75'}
	19%\|█▉ \| 18/96 [24:04<1:40:54, 77.62s/it] 20%\|█▉ \| 19/96 [25:21<1:39:36, 77.62s/it] {'loss': '0.8067', 'grad_norm': '2.438', 'learning_rate': '2e-05', 'ppl': '2.241', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9488', 'tokens/trainable': 14027534, 'tokens/total': 14239739, 'epoch': '0.7917'}
	20%\|█▉ \| 19/96 [25:21<1:39:36, 77.62s/it] 21%\|██ \| 20/96 [26:39<1:38:27, 77.74s/it] {'loss': '0.8145', 'grad_norm': '2.422', 'learning_rate': '2e-05', 'ppl': '2.258', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9524', 'tokens/trainable': 14770394, 'tokens/total': 14988473, 'epoch': '0.8333'}
	21%\|██ \| 20/96 [26:39<1:38:27, 77.74s/it] 22%\|██▏ \| 21/96 [27:57<1:36:59, 77.59s/it] {'loss': '0.7839', 'grad_norm': '2.406', 'learning_rate': '2e-05', 'ppl': '2.19', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9443', 'tokens/trainable': 15499968, 'tokens/total': 15736924, 'epoch': '0.875'}
	22%\|██▏ \| 21/96 [27:57<1:36:59, 77.59s/it] 23%\|██▎ \| 22/96 [29:15<1:35:57, 77.80s/it] {'loss': '0.8035', 'grad_norm': '1.961', 'learning_rate': '2e-05', 'ppl': '2.233', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9450', 'tokens/trainable': 16239694, 'tokens/total': 16485262, 'epoch': '0.9167'}
	23%\|██▎ \| 22/96 [29:15<1:35:57, 77.80s/it] 24%\|██▍ \| 23/96 [30:32<1:34:36, 77.76s/it] {'loss': '0.8221', 'grad_norm': '2.031', 'learning_rate': '2e-05', 'ppl': '2.275', 'memory/max_active (GiB)': '58.05', 'memory/max_allocated (GiB)': '58.05', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9388', 'tokens/trainable': 16968820, 'tokens/total': 17233354, 'epoch': '0.9583'}
	24%\|██▍ \| 23/96 [30:33<1:34:36, 77.76s/it] 25%\|██▌ \| 24/96 [31:46<1:31:53, 76.58s/it] {'loss': '0.7618', 'grad_norm': '2.062', 'learning_rate': '2e-05', 'ppl': '2.142', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '67.02', 'tokens/train_per_sec_per_gpu': '9486', 'tokens/trainable': 17669050, 'tokens/total': 17945280, 'epoch': '1'}
	25%\|██▌ \| 24/96 [31:46<1:31:53, 76.58s/it][2026-08-11 19:36:21,497] [INFO] [axolotl.core.trainers.base.evaluate:464] [PID:31] Running evaluation step...

	0%\| \| 0/90 [00:00<?, ?it/s][A
	2%\|▏ \| 2/90 [00:00<00:31, 2.80it/s][A
	3%\|▎ \| 3/90 [00:00<00:23, 3.72it/s][A
	4%\|▍ \| 4/90 [00:01<00:31, 2.74it/s][A
	6%\|▌ \| 5/90 [00:01<00:33, 2.51it/s][A
	7%\|▋ \| 6/90 [00:02<00:36, 2.28it/s][A
	8%\|▊ \| 7/90 [00:02<00:29, 2.78it/s][A
	9%\|▉ \| 8/90 [00:03<00:33, 2.41it/s][A
	10%\|█ \| 9/90 [00:03<00:31, 2.58it/s][A
	11%\|█ \| 10/90 [00:03<00:27, 2.87it/s][A
	12%\|█▏ \| 11/90 [00:04<00:27, 2.83it/s][A
	13%\|█▎ \| 12/90 [00:04<00:30, 2.58it/s][A
	14%\|█▍ \| 13/90 [00:05<00:33, 2.30it/s][A
	16%\|█▌ \| 14/90 [00:05<00:34, 2.20it/s][A
	17%\|█▋ \| 15/90 [00:06<00:35, 2.09it/s][A
	18%\|█▊ \| 16/90 [00:06<00:28, 2.62it/s][A
	19%\|█▉ \| 17/90 [00:06<00:30, 2.37it/s][A
	20%\|██ \| 18/90 [00:07<00:32, 2.21it/s][A
	21%\|██ \| 19/90 [00:07<00:33, 2.12it/s][A
	22%\|██▏ \| 20/90 [00:08<00:34, 2.06it/s][A
	23%\|██▎ \| 21/90 [00:08<00:26, 2.57it/s][A
	24%\|██▍ \| 22/90 [00:08<00:23, 2.86it/s][A
	26%\|██▌ \| 23/90 [00:09<00:26, 2.48it/s][A
	27%\|██▋ \| 24/90 [00:09<00:28, 2.28it/s][A
	28%\|██▊ \| 25/90 [00:10<00:30, 2.16it/s][A
	29%\|██▉ \| 26/90 [00:10<00:23, 2.75it/s][A
	30%\|███ \| 27/90 [00:10<00:21, 2.93it/s][A
	32%\|███▏ \| 29/90 [00:11<00:18, 3.28it/s][A
	33%\|███▎ \| 30/90 [00:11<00:20, 2.98it/s][A
	34%\|███▍ \| 31/90 [00:12<00:22, 2.59it/s][A
	36%\|███▌ \| 32/90 [00:12<00:22, 2.53it/s][A
	37%\|███▋ \| 33/90 [00:13<00:24, 2.33it/s][A
	38%\|███▊ \| 34/90 [00:13<00:25, 2.19it/s][A
	39%\|███▉ \| 35/90 [00:13<00:22, 2.41it/s][A
	40%\|████ \| 36/90 [00:14<00:23, 2.34it/s][A
	41%\|████ \| 37/90 [00:14<00:20, 2.65it/s][A
	42%\|████▏ \| 38/90 [00:15<00:20, 2.56it/s][A
	43%\|████▎ \| 39/90 [00:15<00:17, 2.93it/s][A
	44%\|████▍ \| 40/90 [00:15<00:16, 3.02it/s][A
	46%\|████▌ \| 41/90 [00:16<00:16, 2.95it/s][A
	47%\|████▋ \| 42/90 [00:16<00:14, 3.27it/s][A
	48%\|████▊ \| 43/90 [00:16<00:16, 2.86it/s][A
	49%\|████▉ \| 44/90 [00:16<00:14, 3.16it/s][A
	50%\|█████ \| 45/90 [00:17<00:14, 3.05it/s][A
	51%\|█████ \| 46/90 [00:17<00:11, 3.74it/s][A
	52%\|█████▏ \| 47/90 [00:17<00:11, 3.79it/s][A
	53%\|█████▎ \| 48/90 [00:18<00:12, 3.46it/s][A
	54%\|█████▍ \| 49/90 [00:18<00:12, 3.37it/s][A
	56%\|█████▌ \| 50/90 [00:18<00:10, 3.72it/s][A
	57%\|█████▋ \| 51/90 [00:19<00:13, 2.89it/s][A
	58%\|█████▊ \| 52/90 [00:19<00:12, 3.00it/s][A
	59%\|█████▉ \| 53/90 [00:19<00:14, 2.57it/s][A
	60%\|██████ \| 54/90 [00:20<00:12, 2.89it/s][A
	61%\|██████ \| 55/90 [00:20<00:14, 2.50it/s][A
	62%\|██████▏ \| 56/90 [00:21<00:12, 2.62it/s][A
	63%\|██████▎ \| 57/90 [00:21<00:10, 3.29it/s][A
	64%\|██████▍ \| 58/90 [00:21<00:08, 3.78it/s][A
	66%\|██████▌ \| 59/90 [00:21<00:10, 2.94it/s][A
	67%\|██████▋ \| 60/90 [00:22<00:11, 2.53it/s][A
	68%\|██████▊ \| 61/90 [00:22<00:10, 2.68it/s][A
	69%\|██████▉ \| 62/90 [00:23<00:11, 2.47it/s][A
	70%\|███████ \| 63/90 [00:23<00:09, 2.80it/s][A
	71%\|███████ \| 64/90 [00:23<00:07, 3.30it/s][A
	72%\|███████▏ \| 65/90 [00:24<00:09, 2.77it/s][A
	73%\|███████▎ \| 66/90 [00:24<00:09, 2.47it/s][A
	74%\|███████▍ \| 67/90 [00:24<00:08, 2.62it/s][A
	76%\|███████▌ \| 68/90 [00:25<00:09, 2.37it/s][A
	77%\|███████▋ \| 69/90 [00:25<00:08, 2.62it/s][A
	78%\|███████▊ \| 70/90 [00:25<00:06, 2.94it/s][A
	79%\|███████▉ \| 71/90 [00:26<00:06, 2.79it/s][A
	80%\|████████ \| 72/90 [00:26<00:05, 3.16it/s][A
	81%\|████████ \| 73/90 [00:26<00:05, 3.31it/s][A
	82%\|████████▏ \| 74/90 [00:27<00:05, 2.72it/s][A
	83%\|████████▎ \| 75/90 [00:27<00:06, 2.42it/s][A
	84%\|████████▍ \| 76/90 [00:28<00:05, 2.69it/s][A
	86%\|████████▌ \| 77/90 [00:28<00:04, 3.10it/s][A
	87%\|████████▋ \| 78/90 [00:28<00:04, 2.83it/s][A
	88%\|████████▊ \| 79/90 [00:29<00:04, 2.46it/s][A
	89%\|████████▉ \| 80/90 [00:29<00:03, 2.70it/s][A
	90%\|█████████ \| 81/90 [00:30<00:03, 2.38it/s][A
	91%\|█████████ \| 82/90 [00:30<00:03, 2.27it/s][A
	92%\|█████████▏\| 83/90 [00:31<00:03, 2.16it/s][A
	93%\|█████████▎\| 84/90 [00:31<00:02, 2.07it/s][A
	94%\|█████████▍\| 85/90 [00:32<00:02, 2.04it/s][A
	96%\|█████████▌\| 86/90 [00:32<00:02, 1.99it/s][A
	97%\|█████████▋\| 87/90 [00:32<00:01, 2.61it/s][A
	98%\|█████████▊\| 88/90 [00:33<00:00, 2.35it/s][A
	99%\|█████████▉\| 89/90 [00:33<00:00, 2.62it/s][A
	100%\|██████████\| 90/90 [00:33<00:00, 2.69it/s][A
	[A{'eval_loss': '0.8244', 'eval_runtime': '35.13', 'eval_samples_per_second': '10.31', 'eval_steps_per_second': '2.591', 'eval_ppl': '2.281', 'memory/max_active (GiB)': '20.04', 'memory/max_allocated (GiB)': '20.04', 'memory/device_reserved (GiB)': '71.9', 'epoch': '1'}
	25%\|██▌ \| 24/96 [32:21<1:31:53, 76.58s/it]
	100%\|██████████\| 90/90 [00:34<00:00, 2.69it/s][A
	[A[2026-08-11 19:36:56,637] [INFO] [axolotl.core.trainers.base._save:928] [PID:31] Saving model checkpoint to ./finetune-model-output/checkpoint-24

	Writing model shards: 0%\| \| 0/1 [00:00<?, ?it/s][A
	Writing model shards: 100%\|██████████\| 1/1 [00:09<00:00, 9.37s/it][A Writing model shards: 100%\|██████████\| 1/1 [00:09<00:00, 9.37s/it]
	[2026-08-11 19:37:30,375] [WARNING] [py.warnings._showwarnmsg:112] [PID:31] /root/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/popen_fork.py:66: DeprecationWarning: This process (pid=31) is multi-threaded, use of fork() may lead to deadlocks in the child.
	self.pid = os.fork()

	26%\|██▌ \| 25/96 [34:17<1:57:04, 98.93s/it] {'loss': '0.645', 'grad_norm': '3.094', 'learning_rate': '2e-05', 'ppl': '1.906', 'memory/max_active (GiB)': '58.06', 'memory/max_allocated (GiB)': '58.06', 'memory/device_reserved (GiB)': '66.16', 'tokens/train_per_sec_per_gpu': '4903', 'tokens/trainable': 18409932, 'tokens/total': 18695242, 'epoch': '1.042'}
	26%\|██▌ \| 25/96 [34:17<1:57:04, 98.93s/it]