Instructions to use CastIronMind/Lissa-1.5B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use CastIronMind/Lissa-1.5B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="CastIronMind/Lissa-1.5B-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("CastIronMind/Lissa-1.5B-Instruct", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use CastIronMind/Lissa-1.5B-Instruct with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/CastIronMind/Lissa-1.5B-Instruct

SGLang

How to use CastIronMind/Lissa-1.5B-Instruct 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 "CastIronMind/Lissa-1.5B-Instruct" \
    --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": "CastIronMind/Lissa-1.5B-Instruct",
		"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 "CastIronMind/Lissa-1.5B-Instruct" \
        --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": "CastIronMind/Lissa-1.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use CastIronMind/Lissa-1.5B-Instruct with Docker Model Runner:
```
docker model run hf.co/CastIronMind/Lissa-1.5B-Instruct
```

Lissa-1.5B-Instruct / tokenizer_config.json

CastIronMind

Add files using upload-large-folder tool

16de7bc verified 6 days ago

Raw History Blame Contribute Delete

2.72 kB

{
  "add_bos_token": false,
  "add_eos_token": false,
  "bos_token": "<bos>",
  "eos_token": "<eos>",
  "pad_token": "<eos>",
  "unk_token": null,
  "clean_up_tokenization_spaces": false,
  "model_max_length": 16384,
  "tokenizer_class": "PreTrainedTokenizerFast",
  "chat_template": "{#-\n  Lissa 1.5B - chat template (ChatML variant with a reasoning block).\n\n  Canonical sequence, as used for the SFT run:\n\n    <bos><im_start><system>{system}<im_end>\n         <im_start><user>{user}<im_end>\n         <im_start><assistant>[<|think|>{think}<|/think|>]{answer}<im_end><eos>\n\n  Notes:\n\n  * <bos> is emitted by this template. The shipped tokenizer.json has a ByteLevel\n    post-processor, not a TemplateProcessing one, so no BOS is added automatically\n    and add_special_tokens has no effect on it.\n\n  * <eos> is emitted only when the conversation ends on an assistant turn, i.e. when\n    the rendered text is a complete training sequence. With add_generation_prompt=True\n    it is not emitted, so the model generates the assistant turn itself.\n\n  * The <|think|> block is optional and is read from the `reasoning_content` field of\n    an assistant message (the `think` field is accepted as an alias). The model is\n    reasoning-first: it opens the block on its own, so it does not need to be\n    prefilled at generation time.\n\n  * <en>, <ru>, </en>, </ru> and <endofprompt> are never emitted. They exist in the\n    vocabulary but were removed from the supervised data.\n\n  * A non-empty system message is required. The model was supervised with one on every\n    example; \"You are a helpful assistant.\" specifically should not be used, it pulls\n    the model towards a generic assistant identity.\n-#}\n{{- '<bos>' -}}\n{%- for message in messages -%}\n    {%- set content = message['content'] if message['content'] is not none else '' -%}\n    {%- if message['role'] == 'system' -%}\n        {{- '<im_start><system>' + content + '<im_end>' -}}\n    {%- elif message['role'] == 'user' -%}\n        {{- '<im_start><user>' + content + '<im_end>' -}}\n    {%- elif message['role'] == 'assistant' -%}\n        {%- set think = message.get('reasoning_content') or message.get('think') -%}\n        {{- '<im_start><assistant>' -}}\n        {%- if think -%}\n            {{- '<|think|>' + think + '<|/think|>' -}}\n        {%- endif -%}\n        {{- content + '<im_end>' -}}\n    {%- else -%}\n        {{- raise_exception('Lissa chat template: unexpected role \"' + message['role'] + '\"') -}}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{- '<im_start><assistant>' -}}\n{%- elif messages and messages[-1]['role'] == 'assistant' -%}\n    {{- '<eos>' -}}\n{%- endif -%}\n"
}

	{
	"add_bos_token": false,
	"add_eos_token": false,
	"bos_token": "<bos>",
	"eos_token": "<eos>",
	"pad_token": "<eos>",
	"unk_token": null,
	"clean_up_tokenization_spaces": false,
	"model_max_length": 16384,
	"tokenizer_class": "PreTrainedTokenizerFast",
	"chat_template": "{#-\n Lissa 1.5B - chat template (ChatML variant with a reasoning block).\n\n Canonical sequence, as used for the SFT run:\n\n <bos><im_start><system>{system}<im_end>\n <im_start><user>{user}<im_end>\n <im_start><assistant>[<\|think\|>{think}<\|/think\|>]{answer}<im_end><eos>\n\n Notes:\n\n * <bos> is emitted by this template. The shipped tokenizer.json has a ByteLevel\n post-processor, not a TemplateProcessing one, so no BOS is added automatically\n and add_special_tokens has no effect on it.\n\n * <eos> is emitted only when the conversation ends on an assistant turn, i.e. when\n the rendered text is a complete training sequence. With add_generation_prompt=True\n it is not emitted, so the model generates the assistant turn itself.\n\n * The <\|think\|> block is optional and is read from the `reasoning_content` field of\n an assistant message (the `think` field is accepted as an alias). The model is\n reasoning-first: it opens the block on its own, so it does not need to be\n prefilled at generation time.\n\n * <en>, <ru>, </en>, </ru> and <endofprompt> are never emitted. They exist in the\n vocabulary but were removed from the supervised data.\n\n * A non-empty system message is required. The model was supervised with one on every\n example; \"You are a helpful assistant.\" specifically should not be used, it pulls\n the model towards a generic assistant identity.\n-#}\n{{- '<bos>' -}}\n{%- for message in messages -%}\n {%- set content = message['content'] if message['content'] is not none else '' -%}\n {%- if message['role'] == 'system' -%}\n {{- '<im_start><system>' + content + '<im_end>' -}}\n {%- elif message['role'] == 'user' -%}\n {{- '<im_start><user>' + content + '<im_end>' -}}\n {%- elif message['role'] == 'assistant' -%}\n {%- set think = message.get('reasoning_content') or message.get('think') -%}\n {{- '<im_start><assistant>' -}}\n {%- if think -%}\n {{- '<\|think\|>' + think + '<\|/think\|>' -}}\n {%- endif -%}\n {{- content + '<im_end>' -}}\n {%- else -%}\n {{- raise_exception('Lissa chat template: unexpected role \"' + message['role'] + '\"') -}}\n {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n {{- '<im_start><assistant>' -}}\n{%- elif messages and messages[-1]['role'] == 'assistant' -%}\n {{- '<eos>' -}}\n{%- endif -%}\n"
	}