Instructions to use sagawa/molscaletransfer-chemlm-2.30m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sagawa/molscaletransfer-chemlm-2.30m with Transformers:

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

pipe = pipeline("text-generation", model="sagawa/molscaletransfer-chemlm-2.30m")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("sagawa/molscaletransfer-chemlm-2.30m")
model = AutoModelForCausalLM.from_pretrained("sagawa/molscaletransfer-chemlm-2.30m")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use sagawa/molscaletransfer-chemlm-2.30m with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sagawa/molscaletransfer-chemlm-2.30m"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sagawa/molscaletransfer-chemlm-2.30m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/sagawa/molscaletransfer-chemlm-2.30m

SGLang

How to use sagawa/molscaletransfer-chemlm-2.30m 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 "sagawa/molscaletransfer-chemlm-2.30m" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sagawa/molscaletransfer-chemlm-2.30m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "sagawa/molscaletransfer-chemlm-2.30m" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sagawa/molscaletransfer-chemlm-2.30m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use sagawa/molscaletransfer-chemlm-2.30m with Docker Model Runner:
```
docker model run hf.co/sagawa/molscaletransfer-chemlm-2.30m
```

molscaletransfer-chemlm-2.30m / args.json

sagawa

Upload 8 files

988048e verified 26 days ago

raw

history blame contribute delete

2.72 kB

	{
	"add_nsp": false,
	"async_worker": true,
	"attention_dropout_checkpoint": false,
	"current_run_id": "",
	"data_loader_type": "dist",
	"dataset_path": "/data2/sagawatatsuya/chemlm_pretraining2/chemlm_pretraining/dataset/data/samples",
	"deepspeed": true,
	"deepspeed_config": "/data2/sagawatatsuya/chemlm_pretraining2/output/2th_size/2th_size-/epoch2048_hf/deepspeed_config.json",
	"deepspeed_transformer_kernel": false,
	"do_validation": true,
	"ds_config": {
	"gradient_clipping": 0.0,
	"steps_per_print": 100,
	"train_batch_size": 4096,
	"train_micro_batch_size_per_gpu": 1024,
	"wall_clock_breakdown": false
	},
	"early_exit_time_marker": 1000000.0,
	"early_stop_eval_loss": 2.1,
	"early_stop_time": 720,
	"finetune_checkpoint_at_end": true,
	"fp16": false,
	"fp16_backend": "ds",
	"fp16_opt": "O2",
	"gelu_checkpoint": false,
	"gradient_accumulation_steps": 4,
	"gradient_clipping": 0.0,
	"job_name": "2th_size",
	"learning_rate": 0.001,
	"load_checkpoint_id": "latest_checkpoint",
	"load_training_checkpoint": "/data2/sagawatatsuya/chemlm_pretraining2/output/2th_size/2th_size-/epoch2048",
	"local_rank": 0,
	"log_throughput_every": 20,
	"lr": 0.001,
	"max_predictions_per_seq": 77,
	"max_steps": 9223372036854775807,
	"max_steps_per_epoch": 9223372036854775807,
	"model_config": {
	"attention_probs_dropout_prob": 0.1,
	"encoder_ln_mode": "pre-ln",
	"fused_linear_layer": true,
	"hidden_act": "gelu",
	"hidden_dropout_prob": 0.1,
	"hidden_size": 192,
	"initializer_range": 0.02,
	"intermediate_size": 768,
	"layer_norm_type": "apex",
	"layernorm_embedding": false,
	"max_position_embeddings": 512,
	"num_attention_heads": 3,
	"num_hidden_layers": 5,
	"sparse_mask_prediction": true,
	"type_vocab_size": 2,
	"vocab_size": 2362
	},
	"model_type": "bert-mlm",
	"no_nsp": true,
	"normalize_invertible": false,
	"num_epochs": 2048,
	"num_epochs_between_checkpoints": 1,
	"num_workers": 12,
	"output_dir": "/data2/sagawatatsuya/chemlm_pretraining2/output/2th_size",
	"prescale_gradients": false,
	"print_steps": 100,
	"project_name": "chemlm_pretraining_2th_size",
	"scale_cnt_limit": 100,
	"seed": 42,
	"steps_per_print": 100,
	"stochastic_mode": false,
	"tokenizer_name": "ibm-research/MoLFormer-XL-both-10pct",
	"total_training_time": 1000000.0,
	"train_batch_size": 4096,
	"train_micro_batch_size_per_gpu": 1024,
	"use_early_stopping": false,
	"validation_begin_proportion": 0.05,
	"validation_end_proportion": 0.01,
	"validation_epochs": 1,
	"validation_epochs_begin": 1,
	"validation_epochs_end": 1,
	"validation_micro_batch": 1024,
	"vocab_size": 2368,
	"wall_clock_breakdown": false
	}