Instructions to use Jboadu/new-gaia with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Jboadu/new-gaia with Transformers:

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

pipe = pipeline("text-generation", model="Jboadu/new-gaia")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Jboadu/new-gaia")
model = AutoModelForCausalLM.from_pretrained("Jboadu/new-gaia")
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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Jboadu/new-gaia with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Jboadu/new-gaia

SGLang

How to use Jboadu/new-gaia 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 "Jboadu/new-gaia" \
    --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": "Jboadu/new-gaia",
		"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 "Jboadu/new-gaia" \
        --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": "Jboadu/new-gaia",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Jboadu/new-gaia with Docker Model Runner:
```
docker model run hf.co/Jboadu/new-gaia
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

See axolotl config

axolotl version: 0.12.0

base_model: Jboadu/test-model-2-pretrain
tokenizer_type: AutoTokenizer
model_type: AutoModelForCausalLM
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
- path: axolotl_correction_conversations_GAIA_Raw_Training_Data.json
  type: input_output
- path: factual_sft_completion/combined_all_0.jsonl
  type: completion
- path: factual_sft_completion/combined_all_2.jsonl
  type: completion
- path: factual_sft_completion/combined_all_6.jsonl
  type: completion
- path: factual_sft_completion/combined_all_4.jsonl
  type: completion
- path: factual_sft_completion/combined_all_3.jsonl
  type: completion
- path: factual_sft_completion/combined_all_1.jsonl
  type: completion
- path: factual_sft_completion/combined_all_5.jsonl
  type: completion
- path: factual_sft_completion/combined_all_7.jsonl
  type: completion
- path: generic_sft_completion/Augmentoolkit-Augmentoolkit-Capybara-2point5mil-Thoughts_300000.jsonl
  type: completion
- path: generic_sft_completion/Augmentoolkit-Openthoughts-100mil-DifferentFormat_600000.jsonl
  type: completion
- path: generic_sft_completion/Augmentoolkit-Augmentoolkit-Generic-Grabbag-Thoughts_400000.jsonl
  type: completion
- path: generic_sft_completion/Augmentoolkit-Augmentoolkit-LMsys-800k-Thoughts_200000.jsonl
  type: completion
dataset_prepared_path: last_finetune_prepared
output_dir: ./finetune-model-output
seed: 1337
sequence_len: 5000
sample_packing: true
pad_to_sequence_len: false
shuffle_merged_datasets: true
gradient_accumulation_steps: 75
micro_batch_size: 2
eval_batch_size: 4
num_epochs: 5
optimizer: paged_adamw_8bit
lr_scheduler: constant
learning_rate: 2.0e-05
noisy_embedding_alpha: 5
weight_decay: 0
train_on_inputs: false
group_by_length: false
bf16: true
fp16: false
tf32: false
gradient_checkpointing: true
logging_steps: 1
xformers_attention: false
flash_attention: true
chat_template: chatml
auto_resume_from_checkpoints: false
warmup_ratio: 0.1
evals_per_epoch: 1
val_set_size: 0.04
saves_per_epoch: 1
eval_sample_packing: false
save_total_limit: 2
special_tokens:
  pad_token: <unk>
use_liger_kernel: true
plugins:
- axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
liger_layer_norm: true
liger_fused_linear_cross_entropy: true
sequence_length: 10000
wandb_project: test-project
wandb_entity: ''
wandb_watch: ''
wandb_run_id: ''
wandb_log_model: ''
hub_model_id: Jboadu/new-gaia
hub_strategy: all_checkpoints

new-gaia

This model is a fine-tuned version of Jboadu/test-model-2-pretrain on the axolotl_correction_conversations_GAIA_Raw_Training_Data.json, the factual_sft_completion/combined_all_0.jsonl, the factual_sft_completion/combined_all_2.jsonl, the factual_sft_completion/combined_all_6.jsonl, the factual_sft_completion/combined_all_4.jsonl, the factual_sft_completion/combined_all_3.jsonl, the factual_sft_completion/combined_all_1.jsonl, the factual_sft_completion/combined_all_5.jsonl, the factual_sft_completion/combined_all_7.jsonl, the generic_sft_completion/Augmentoolkit-Augmentoolkit-Capybara-2point5mil-Thoughts_300000.jsonl, the generic_sft_completion/Augmentoolkit-Openthoughts-100mil-DifferentFormat_600000.jsonl, the generic_sft_completion/Augmentoolkit-Augmentoolkit-Generic-Grabbag-Thoughts_400000.jsonl and the generic_sft_completion/Augmentoolkit-Augmentoolkit-LMsys-800k-Thoughts_200000.jsonl datasets. It achieves the following results on the evaluation set:

Loss: 0.5799
Memory/max Mem Active(gib): 31.49
Memory/max Mem Allocated(gib): 31.49
Memory/device Mem Reserved(gib): 33.38

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 2
eval_batch_size: 4
seed: 1337
gradient_accumulation_steps: 75
total_train_batch_size: 150
optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: constant
lr_scheduler_warmup_steps: 2
training_steps: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Mem Active(gib)	Mem Allocated(gib)	Mem Reserved(gib)
No log	0	0	1.4924	19.79	19.79	23.71
0.8381	0.9585	4	0.7330	31.49	31.49	33.38
0.5844	1.7188	8	0.6324	31.49	31.49	33.38
0.4746	2.4792	12	0.5766	31.49	31.49	33.38
0.3431	3.4792	16	0.5799	31.49	31.49	33.38