Text Generation
Transformers
PyTorch
Safetensors
English
llama
text-generation-inference
unsloth
trl
conversational
Instructions to use EpistemeAI/ReasoningCore-3B-R01 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use EpistemeAI/ReasoningCore-3B-R01 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="EpistemeAI/ReasoningCore-3B-R01") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("EpistemeAI/ReasoningCore-3B-R01") model = AutoModelForCausalLM.from_pretrained("EpistemeAI/ReasoningCore-3B-R01", 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=40) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use EpistemeAI/ReasoningCore-3B-R01 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "EpistemeAI/ReasoningCore-3B-R01" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "EpistemeAI/ReasoningCore-3B-R01", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/EpistemeAI/ReasoningCore-3B-R01
- SGLang
How to use EpistemeAI/ReasoningCore-3B-R01 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 "EpistemeAI/ReasoningCore-3B-R01" \ --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": "EpistemeAI/ReasoningCore-3B-R01", "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 "EpistemeAI/ReasoningCore-3B-R01" \ --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": "EpistemeAI/ReasoningCore-3B-R01", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Unsloth Studio
How to use EpistemeAI/ReasoningCore-3B-R01 with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for EpistemeAI/ReasoningCore-3B-R01 to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for EpistemeAI/ReasoningCore-3B-R01 to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for EpistemeAI/ReasoningCore-3B-R01 to start chatting
Load model with FastModel
pip install unsloth from unsloth import FastModel model, tokenizer = FastModel.from_pretrained( model_name="EpistemeAI/ReasoningCore-3B-R01", max_seq_length=2048, ) - Docker Model Runner
How to use EpistemeAI/ReasoningCore-3B-R01 with Docker Model Runner:
docker model run hf.co/EpistemeAI/ReasoningCore-3B-R01
Update README.md
Browse files
README.md
CHANGED
|
@@ -11,11 +11,9 @@ language:
|
|
| 11 |
- en
|
| 12 |
---
|
| 13 |
|
| 14 |
-
#
|
| 15 |
-
|
| 16 |
-
**ReasomingCore‑3B** is a multilingual, reasoning‑enhanced large language model developed by EpitemeAI. Pretrained on vast amounts of publicly available data and instruction‑tuned to excel at nuanced reasoning, dialogue management, retrieval, and summarization tasks, it often outperforms many current open source and proprietary conversational models on a range of industry benchmarks.
|
| 17 |
-
Supervised fine tuned with ReasoningCore-3B-0 with reasoning dataset.
|
| 18 |
|
|
|
|
| 19 |
|
| 20 |
---
|
| 21 |
|
|
@@ -23,11 +21,11 @@ Supervised fine tuned with ReasoningCore-3B-0 with reasoning dataset.
|
|
| 23 |
|
| 24 |
- **Model Developer:** EpitemeAI
|
| 25 |
- **Model Architecture:**
|
| 26 |
-
|
| 27 |
|
| 28 |
| | Training Data | Params | Input Modalities | Output Modalities | Context Length | GQA | Shared Embeddings | Token Count | Knowledge Cutoff |
|
| 29 |
|--------------------------------|--------------------------------------------------|--------|-----------------------|------------------------------|----------------|-----|-------------------|----------------|-------------------|
|
| 30 |
-
| **
|
| 31 |
|
| 32 |
- **Supported Languages:**
|
| 33 |
Officially supports English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai. While the pretraining included a broader range of languages, additional languages can be fine‑tuned in compliance with the community license and acceptable use policies.
|
|
@@ -55,12 +53,9 @@ Supervised fine tuned with ReasoningCore-3B-0 with reasoning dataset.
|
|
| 55 |
|
| 56 |
## How to Use
|
| 57 |
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
### Use with Transformers
|
| 61 |
-
|
| 62 |
|
| 63 |
-
Use system prompt
|
| 64 |
```bash
|
| 65 |
SYSTEM_PROMPT = """
|
| 66 |
Respond in the following format:
|
|
@@ -73,6 +68,8 @@ Respond in the following format:
|
|
| 73 |
"""
|
| 74 |
```
|
| 75 |
|
|
|
|
|
|
|
| 76 |
Ensure you have transformers version 4.43.0 or later installed:
|
| 77 |
|
| 78 |
```bash
|
|
@@ -90,24 +87,6 @@ pipe = pipeline(
|
|
| 90 |
)
|
| 91 |
print(pipe("The secret to effective reasoning is"))
|
| 92 |
```
|
| 93 |
-
|
| 94 |
-
### Test output 'aha moment'
|
| 95 |
-
prompt: Continue the fibonnaci sequence: 1, 1, 2, 3, 5, 8,
|
| 96 |
-
|
| 97 |
-
```python
|
| 98 |
-
Okay, so I need to continue the Fibonacci sequence. Let me start by recalling what the Fibonacci sequence is. It's a series of numbers where each number is the sum of the two preceding ones, usually starting with 0 and 1. Wait, but the problem says it starts with 1, 1, 2, 3, 5, 8. Hmm, that's a bit different. Maybe it's a variation or something. Let me check.
|
| 99 |
-
|
| 100 |
-
Wait, the standard Fibonacci sequence starts with 0 and 1, but this one starts with 1 and 1. So maybe it's a shifted version? Let me see. If it starts with 1, 1, then the next number would be 1 + 1 = 2, which is correct. Then 2 + 1 = 3, 3 + 2 = 5, 5 + 3 = 8. So yes, that's the same as the standard sequence. So the next numbers would be 8 + 5 = 13, 13 + 8 = 21, 21 + 13 = 34, 34 + 21 = 55, 55 + 34 = 89, 89 + 55 = 144, and so on.
|
| 101 |
-
|
| 102 |
-
But wait, the problem says it's the Fibonacci sequence. So maybe I should just stick with the standard sequence. Let me list them out again:
|
| 103 |
-
|
| 104 |
-
1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144,...
|
| 105 |
-
|
| 106 |
-
Yes, that seems right. So the next number would be 89 + 55 = 144. Then 144 + 89 = 233, 233 + 144 = 377, 377 + 233 = 610, 610 + 377 = 987, 987 + 610 = 1597, 1597 + 987 = 2584, 2584 + 1597 = 4181, 4181 + 2584 = 6765, 6765 + 4181 = 10946, 10946 + 6765 = 17711, 17711 + 10946 = 28657, 28657 + 17711 = 46368, 46368 + 28657 = 75025, 75025 + 46368 = 121393, 121393 + 75025 = 196418, 196418 + 121393 = 317811, 317811 + 196418 = 514229, 514229 + 317811 = 832040, 832040 + 514229 = 1346869, 1346869 + 832040 = 2176909, 2176909 + 1346869 = 3523778, 3523778 + 217
|
| 107 |
-
```
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
## For Mathematical problems
|
| 112 |
Please use "Please reason step by step, and put your final answer within \boxed{}" in system prompt
|
| 113 |
|
|
@@ -117,7 +96,7 @@ Please use "Please reason step by step, and put your final answer within \boxed{
|
|
| 117 |
### Responsible Deployment
|
| 118 |
|
| 119 |
#### Approach:
|
| 120 |
-
- **
|
| 121 |
|
| 122 |
#### System‑Level Safety:
|
| 123 |
- The model is designed to be deployed as part of a broader system that implements safety measures (e.g., Prompt Guard, Code Shield) to ensure outputs remain safe even under adversarial conditions.
|
|
@@ -163,7 +142,7 @@ Please use "Please reason step by step, and put your final answer within \boxed{
|
|
| 163 |
### Ethical Considerations and Limitations
|
| 164 |
|
| 165 |
#### Core Values:
|
| 166 |
-
- **
|
| 167 |
|
| 168 |
#### Testing and Limitations:
|
| 169 |
- Despite extensive testing across diverse scenarios, the model may occasionally produce inaccurate, biased, or objectionable outputs. Developers must perform additional safety testing and integrate further safeguards as needed.
|
|
@@ -177,7 +156,7 @@ Please use "Please reason step by step, and put your final answer within \boxed{
|
|
| 177 |
|
| 178 |
### Conclusion
|
| 179 |
|
| 180 |
-
**
|
| 181 |
|
| 182 |
For further details, questions, or feedback, please email episteme.ai@proton.me
|
| 183 |
|
|
|
|
| 11 |
- en
|
| 12 |
---
|
| 13 |
|
| 14 |
+
# ReasoningCore‑3B-0
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
+
**ReasoningCore‑3B** is a multilingual, reasoning‑enhanced large language model developed by EpitemeAI. Pretrained on vast amounts of publicly available data and instruction‑tuned to excel at nuanced reasoning, dialogue management, retrieval, and summarization tasks, it often outperforms many current open source and proprietary conversational models on a range of industry benchmarks. Fine tuned with reasoning dataset.
|
| 17 |
|
| 18 |
---
|
| 19 |
|
|
|
|
| 21 |
|
| 22 |
- **Model Developer:** EpitemeAI
|
| 23 |
- **Model Architecture:**
|
| 24 |
+
ReasoningCore‑3B is an auto‑regressive language model built on an optimized transformer architecture. It incorporates specialized reasoning pathways and has been fine‑tuned using both supervised learning and reinforcement learning with human feedback (RLHF) to align with human expectations for clarity, accuracy, and safety in complex tasks.
|
| 25 |
|
| 26 |
| | Training Data | Params | Input Modalities | Output Modalities | Context Length | GQA | Shared Embeddings | Token Count | Knowledge Cutoff |
|
| 27 |
|--------------------------------|--------------------------------------------------|--------|-----------------------|------------------------------|----------------|-----|-------------------|----------------|-------------------|
|
| 28 |
+
| **ReasoningCore‑3B (text only)** | A new mix of publicly available online data. | 3B | Multilingual Text | Multilingual Text and code | 128k | Yes | Yes | Up to 9T tokens | December 2023 |
|
| 29 |
|
| 30 |
- **Supported Languages:**
|
| 31 |
Officially supports English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai. While the pretraining included a broader range of languages, additional languages can be fine‑tuned in compliance with the community license and acceptable use policies.
|
|
|
|
| 53 |
|
| 54 |
## How to Use
|
| 55 |
|
| 56 |
+
ReasoningCore‑3B can be integrated using popular machine learning frameworks. Two primary methods are provided:
|
|
|
|
|
|
|
|
|
|
| 57 |
|
| 58 |
+
## Use system prompt
|
| 59 |
```bash
|
| 60 |
SYSTEM_PROMPT = """
|
| 61 |
Respond in the following format:
|
|
|
|
| 68 |
"""
|
| 69 |
```
|
| 70 |
|
| 71 |
+
### Use with Transformers
|
| 72 |
+
|
| 73 |
Ensure you have transformers version 4.43.0 or later installed:
|
| 74 |
|
| 75 |
```bash
|
|
|
|
| 87 |
)
|
| 88 |
print(pipe("The secret to effective reasoning is"))
|
| 89 |
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
## For Mathematical problems
|
| 91 |
Please use "Please reason step by step, and put your final answer within \boxed{}" in system prompt
|
| 92 |
|
|
|
|
| 96 |
### Responsible Deployment
|
| 97 |
|
| 98 |
#### Approach:
|
| 99 |
+
- **ReasoningCore‑3B** is a foundational technology that includes built‑in safety guardrails. Developers are encouraged to integrate additional safeguards tailored to their specific applications.
|
| 100 |
|
| 101 |
#### System‑Level Safety:
|
| 102 |
- The model is designed to be deployed as part of a broader system that implements safety measures (e.g., Prompt Guard, Code Shield) to ensure outputs remain safe even under adversarial conditions.
|
|
|
|
| 142 |
### Ethical Considerations and Limitations
|
| 143 |
|
| 144 |
#### Core Values:
|
| 145 |
+
- **ReasoningCore‑3B** is built on the values of openness, inclusivity, and helpfulness. It is designed to respect user autonomy and foster free thought and expression while mitigating potential harm.
|
| 146 |
|
| 147 |
#### Testing and Limitations:
|
| 148 |
- Despite extensive testing across diverse scenarios, the model may occasionally produce inaccurate, biased, or objectionable outputs. Developers must perform additional safety testing and integrate further safeguards as needed.
|
|
|
|
| 156 |
|
| 157 |
### Conclusion
|
| 158 |
|
| 159 |
+
**ReasoningCore‑3B** represents a significant advancement in multilingual, reasoning‑enhanced language models. Optimized for tasks requiring deep reasoning, contextual understanding, and safe, helpful interactions, it offers a powerful tool for both commercial and research applications. We invite developers and researchers to explore its capabilities and contribute to building secure, innovative AI systems.
|
| 160 |
|
| 161 |
For further details, questions, or feedback, please email episteme.ai@proton.me
|
| 162 |
|