Instructions to use dirac-run/ec-1.5b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dirac-run/ec-1.5b with Transformers:

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

pipe = pipeline("text-generation", model="dirac-run/ec-1.5b")
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("dirac-run/ec-1.5b")
model = AutoModelForCausalLM.from_pretrained("dirac-run/ec-1.5b", 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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use dirac-run/ec-1.5b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dirac-run/ec-1.5b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dirac-run/ec-1.5b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dirac-run/ec-1.5b

SGLang

How to use dirac-run/ec-1.5b 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 "dirac-run/ec-1.5b" \
    --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": "dirac-run/ec-1.5b",
		"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 "dirac-run/ec-1.5b" \
        --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": "dirac-run/ec-1.5b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use dirac-run/ec-1.5b with Docker Model Runner:
```
docker model run hf.co/dirac-run/ec-1.5b
```

ec-1.5b / training.json

dirac-run

Publish selected EasyCommand checkpoint

59b3637 verified 5 days ago

Raw History Blame Contribute Delete

2.88 kB

	{
	"base_model": {
	"repository": "Qwen/Qwen2.5-Coder-1.5B-Instruct",
	"revision": "2e1fd397ee46e1388853d2af2c993145b0f1098a",
	"parameter_count": 1543714304
	},
	"selected_checkpoint": "A3",
	"parent_checkpoint_step": 14833,
	"parent_training": {
	"seed": 42,
	"effective_batch": 32,
	"rank": 32,
	"alpha": 64,
	"dropout": 0.05,
	"target_modules": [
	"q_proj",
	"k_proj",
	"v_proj",
	"o_proj",
	"gate_proj",
	"up_proj",
	"down_proj"
	],
	"rows": 474635,
	"peak_learning_rate": 0.0001,
	"updates": 14833,
	"warmup_updates": 742,
	"optimizer": "AdamW",
	"betas": [
	0.9,
	0.999
	],
	"epsilon": 1e-08,
	"weight_decay": 0.0,
	"schedule": "linear warmup then cosine decay",
	"base_dtype": "bfloat16",
	"trainable_adapter_dtype": "float32",
	"max_sequence_tokens": 2048,
	"assistant_only_loss": true,
	"length_bucketed": true,
	"prompt_file": "training-system-prompt.txt",
	"prompt_sha256": "80e46c8e196e108ea50420d1c7b0b28d9aa07a2df1ccbf49ebd63d466c4a37f5",
	"epochs": 1,
	"example_order": "shuffled length buckets; one complete weighted epoch"
	},
	"continuation": {
	"seed": 23042,
	"effective_batch": 32,
	"rank": 32,
	"alpha": 64,
	"dropout": 0.05,
	"target_modules": [
	"q_proj",
	"k_proj",
	"v_proj",
	"o_proj",
	"gate_proj",
	"up_proj",
	"down_proj"
	],
	"rows": 2820,
	"peak_learning_rate": 5e-06,
	"updates": 200,
	"warmup_updates": 10,
	"optimizer": "AdamW",
	"betas": [
	0.9,
	0.999
	],
	"epsilon": 1e-08,
	"weight_decay": 0.0,
	"schedule": "linear warmup then cosine decay",
	"base_dtype": "bfloat16",
	"trainable_adapter_dtype": "float32",
	"max_sequence_tokens": 2048,
	"assistant_only_loss": true,
	"length_bucketed": false,
	"prompt_file": "system-prompt.txt",
	"prompt_sha256": "3a9028d5aebb73c3ed7363e63eb3e751689218775ea5572ab0dc6807e522238b",
	"repair_mix_fraction": 0.5,
	"example_order": "seeded family-aware shuffle; repair/replay cohorts without replacement within cycles"
	},
	"training_data_release": "dirac-run/ec-training-data",
	"data_release_is_deduplicated_union_not_exact_weighted_epoch": true,
	"training_examples_from_other_repairs_also_in_data_release": true,
	"full_fidelity_artifacts_in_this_repository": true,
	"merged_dtype": "bfloat16",
	"adapter_dtype": "float32",
	"adapter_subfolder": "adapter",
	"optimizer_and_rng_state_included": false,
	"merge": "FP32(original BF16 base) + (alpha/r) * FP32(B @ A), then BF16 rounding",
	"gguf_repository": "dirac-run/ec-1.5b-gguf",
	"source_adapter_sha256": "3ea9ba05f368b8be44ea02e7f6618391ecae8248ea9c550e72397957a1038042",
	"source_fp32_merge_sha256": "4f346d07a74b320a88ef06fa09e614ce5d071e4f71f68076c3a79575575ad7d2"
	}