--- title: cogito-v2-preview-llama-405B canonical_url: "https://www.modelscope.cn/models/unsloth/cogito-v2-preview-llama-405B" md_url: "https://www.modelscope.cn/models/unsloth/cogito-v2-preview-llama-405B.md" repository: unsloth/cogito-v2-preview-llama-405B last_updated: 2025-07-31 license: llama3.1 pipeline_tag: text-generation tasks: - text-generation model_type: - llama architectures: - LlamaForCausalLM base_model: - deepcogito/cogito-v2-preview-llama-405B base_model_relation: finetune parameters: 405.9B tensor_type: - BF16 library_name: - pytorch - transformer - safetensors frameworks: - pytorch inference_backends: - "deploy_task text/emb" - "lmdeploy 0.9.1" - "lmdeploy_turbomind 0.9.1" - "sglang 0.5.2" - "vllm 0.9.2" downloads: 142 stars: 0 tags: - unsloth --- # cogito-v2-preview-llama-405B > cogito-v2-preview-llama-405B - unsloth 在 ModelScope 开源的模型。[!NOTE] Includes Unsloth chat template fixes! For llama.cpp, use --jinja unsloth/cogito-v2-preview-llama-405B 是 ModelScope 魔搭社区上的 405.9B 参数text-generation模型，采用 llama3.1 许可，基于 deepcogito/cogito-v2-preview-llama-405B 构建，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。 - **Repository**: unsloth/cogito-v2-preview-llama-405B - **License**: llama3.1 - **Tasks**: text-generation - **Parameters**: 405.9B - **Base model**: deepcogito/cogito-v2-preview-llama-405B - **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2 - **Tags**: unsloth - **Downloads**: 142 - **Stars**: 0 - **Last updated**: 2025-07-31 Source: https://www.modelscope.cn/models/unsloth/cogito-v2-preview-llama-405B --- > [!NOTE] > Includes Unsloth **chat template fixes**!
For `llama.cpp`, use `--jinja` >

Unsloth Dynamic 2.0 achieves superior accuracy & outperforms other leading quants.

Logo

# Cogito v2 preview - 405B [Blog Post](https://www.deepcogito.com/research/cogito-v2-preview) The Cogito v2 LLMs are instruction tuned generative models. All models are released under an open license for commercial use. - Cogito v2 models are hybrid reasoning models. Each model can answer directly (standard LLM), or self-reflect before answering (like reasoning models). - The LLMs are trained using **Iterated Distillation and Amplification (IDA)** - an scalable and efficient alignment strategy for superintelligence using iterative self-improvement. - The models have been optimized for coding, STEM, instruction following and general helpfulness, and have significantly higher multilingual, coding and tool calling capabilities than size equivalent counterparts. - In both standard and reasoning modes, Cogito v2-preview models outperform their size equivalent counterparts on common industry benchmarks. - This model is trained in over 30 languages and supports a context length of 128k. # Evaluations Here is the model performance on some standard industry benchmarks:

Logo

For detailed evaluations, please refer to the [Blog Post](https://www.deepcogito.com/research/cogito-v2-preview). # Usage Here is a snippet below for usage with Transformers: ```python import transformers import torch model_id = "deepcogito/cogito-v2-preview-llama-405B" pipeline = transformers.pipeline( "text-generation", model=model_id, model_kwargs={"torch_dtype": torch.bfloat16}, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Give me a short introduction to LLMs."}, ] outputs = pipeline( messages, max_new_tokens=512, ) print(outputs[0]["generated_text"][-1]) ``` ## Implementing extended thinking - By default, the model will answer in the standard mode. - To enable thinking, you can do any one of the two methods: - Set `enable_thinking=True` while applying the chat template. - Add a specific system prompt, along with prefilling the response with "\\n". **NOTE: Unlike Cogito v1 models, we initiate the response with "\\n" at the beginning of every output when reasoning is enabled. This is because hybrid models can be brittle at times (<0.1% of the cases), and adding a "\\n" ensures that the model does indeed respect thinking.** ### Method 1 - Set enable_thinking=True in the tokenizer If you are using Huggingface tokenizers, then you can simply use add the argument `enable_thinking=True` to the tokenization (this option is added to the chat template). Here is an example - ```python from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "deepcogito/cogito-v2-preview-llama-405B" model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto" ) tokenizer = AutoTokenizer.from_pretrained(model_name) prompt = "Give me a short introduction to LLMs." messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": prompt} ] text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True, enable_thinking=True ) model_inputs = tokenizer([text], return_tensors="pt").to(model.device) generated_ids = model.generate( **model_inputs, max_new_tokens=512 ) generated_ids = [ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids) ] response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] print(response) ``` ### Method 2 - Add a specific system prompt, along with prefilling the response with "\\n". To enable thinking using this method, you need to do two parts - Step 1 - Simply use this in the system prompt `system_instruction = 'Enable deep thinking subroutine.'` If you already have a system_instruction, then use `system_instruction = 'Enable deep thinking subroutine.' + '\n\n' + system_instruction`. Step 2 - Prefil the response with the tokens `"\n"`. Here is an example - ```python import transformers import torch model_name = "deepcogito/cogito-v2-preview-llama-405B" model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto" ) tokenizer = AutoTokenizer.from_pretrained(model_name) # Step 1 - Add deep thinking instruction. DEEP_THINKING_INSTRUCTION = "Enable deep thinking subroutine." messages = [ {"role": "system", "content": DEEP_THINKING_INSTRUCTION}, {"role": "user", "content": "Write a bash script that takes a matrix represented as a string with format '[1,2],[3,4],[5,6]' and prints the transpose in the same format."}, ] text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) # Step 2 - Prefill response with "\n". text += "\n" # Now, continue as usual. model_inputs = tokenizer([text], return_tensors="pt").to(model.device) generated_ids = model.generate( **model_inputs, max_new_tokens=512 ) generated_ids = [ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids) ] response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] print(response) ``` Similarly, if you have a system prompt, you can append the `DEEP_THINKING_INSTRUCTION` to the beginning in this way - ```python DEEP_THINKING_INSTRUCTION = "Enable deep thinking subroutine." system_prompt = "Reply to each prompt with only the actual code - no explanations." prompt = "Write a bash script that takes a matrix represented as a string with format '[1,2],[3,4],[5,6]' and prints the transpose in the same format." messages = [ {"role": "system", "content": DEEP_THINKING_INSTRUCTION + '\n\n' + system_prompt}, {"role": "user", "content": prompt} ] ``` # Tool Calling Cogito models support tool calling (single, parallel, multiple and parallel_multiple) both in standard and extended thinking mode. Here is a snippet - ```python # First, define a tool def get_current_temperature(location: str) -> float: """ Get the current temperature at a location. Args: location: The location to get the temperature for, in the format "City, Country" Returns: The current temperature at the specified location in the specified units, as a float. """ return 22. # A real function should probably actually get the temperature! # Next, create a chat and apply the chat template messages = [ {"role": "user", "content": "Hey, what's the temperature in Paris right now?"} ] model_inputs = tokenizer.apply_chat_template(messages, tools=[get_current_temperature], add_generation_prompt=True) text = tokenizer.apply_chat_template(messages, tools=[get_current_temperature], add_generation_prompt=True, tokenize=False) inputs = tokenizer(text, return_tensors="pt", add_special_tokens=False).to(model.device) outputs = model.generate(**inputs, max_new_tokens=512) output_text = tokenizer.batch_decode(outputs)[0][len(text):] print(output_text) ``` This will result in the output - ``` {"name": "get_current_temperature", "arguments": {"location": "Paris, France"}} <|eot_id|> ``` You can then generate text from this input as normal. If the model generates a tool call, you should add it to the chat like so: ```python tool_call = {"name": "get_current_temperature", "arguments": {"location": "Paris, France"}} messages.append({"role": "assistant", "tool_calls": [{"type": "function", "function": tool_call}]}) ``` and then call the tool and append the result, with the `tool` role, like so: ```python messages.append({"role": "tool", "name": "get_current_temperature", "content": "22.0"}) ``` After that, you can `generate()` again to let the model use the tool result in the chat: ```python text = tokenizer.apply_chat_template(messages, tools=[get_current_temperature], add_generation_prompt=True, tokenize=False) inputs = tokenizer(text, return_tensors="pt", add_special_tokens=False).to(model.device) outputs = model.generate(**inputs, max_new_tokens=512) output_text = tokenizer.batch_decode(outputs)[0][len(text):] ``` This should result in the string - ``` 'The current temperature in Paris is 22.0 degrees.<|eot_id|>' ``` ## License This repository and the model weights are licensed under the [Llama 3.1 Community License Agreement](https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/LICENSE) (Llama models' default license agreement). ## Contact If you would like to reach out to our team, send an email to [contact@deepcogito.com](contact@deepcogito.com).