Triangle104
commited on
Update README.md
Browse files
README.md
CHANGED
@@ -121,6 +121,61 @@ model-index:
|
|
121 |
This model was converted to GGUF format from [`Daemontatox/Sphinx2.0`](https://huggingface.co/Daemontatox/Sphinx2.0) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
122 |
Refer to the [original model card](https://huggingface.co/Daemontatox/Sphinx2.0) for more details on the model.
|
123 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
124 |
## Use with llama.cpp
|
125 |
Install llama.cpp through brew (works on Mac and Linux)
|
126 |
|
|
|
121 |
This model was converted to GGUF format from [`Daemontatox/Sphinx2.0`](https://huggingface.co/Daemontatox/Sphinx2.0) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
122 |
Refer to the [original model card](https://huggingface.co/Daemontatox/Sphinx2.0) for more details on the model.
|
123 |
|
124 |
+
---
|
125 |
+
Model details:
|
126 |
+
-
|
127 |
+
phinx: The Apex of Logical Deduction and Chain-of-Thought Reasoning
|
128 |
+
|
129 |
+
Developed by: Daemontatox
|
130 |
+
License: Apache-2.0
|
131 |
+
Base Model: Fine-tuned from unsloth/qwen2.5-14b-instruct-bnb-4bit
|
132 |
+
Accelerated by: Unsloth Framework
|
133 |
+
TRL-Optimized: Integrated with Huggingface's TRL library for enhanced performance in logical reasoning.
|
134 |
+
|
135 |
+
Unveiling Sphinx: Master of Reasoned Thought
|
136 |
+
|
137 |
+
Sphinx is a cutting-edge Long Chain-of-Thought (CoT) reasoning model meticulously crafted to unravel complex challenges requiring rigorous logical analysis. Built upon the robust foundation of the Qwen2.5 architecture, Sphinx excels at constructing coherent, step-by-step thought processes, providing unparalleled insight into its reasoning and ensuring clarity in its conclusions.
|
138 |
+
|
139 |
+
"Where complexity yields to logical clarity."
|
140 |
+
|
141 |
+
Core Strengths: Reasoning, Logic, and CoT
|
142 |
+
|
143 |
+
Unrivaled Chain-of-Thought (CoT) Mastery: Engineered for dissecting intricate problems, Sphinx meticulously constructs each step of its reasoning, offering a transparent and verifiable pathway to the solution.
|
144 |
+
Deep Logical Reasoning Capabilities: Sphinx is adept at navigating complex logical structures, drawing valid inferences and forming sound conclusions through multi-layered analysis.
|
145 |
+
Exceptional Reasoning Fidelity: Fine-tuned to maintain the highest standards of logical consistency, Sphinx delivers outputs that are not only correct but also demonstrably well-reasoned.
|
146 |
+
Efficient Long-Context Reasoning: Leveraging the power of Unsloth, Sphinx processes extensive information efficiently, maintaining logical coherence across extended reasoning chains.
|
147 |
+
Explainable AI through Transparent Logic: Sphinx's inherent CoT approach provides explicit and understandable reasoning, making its decision-making process transparent and trustworthy.
|
148 |
+
|
149 |
+
Model Architecture and Fine-tuning for Logical Prowess
|
150 |
+
Architectural Foundation
|
151 |
+
|
152 |
+
Base Model: Qwen2.5-14B - Renowned for its strong general language understanding, forming a solid basis for specialized reasoning.
|
153 |
+
Parameters: 14 billion - Providing the capacity to model intricate reasoning patterns.
|
154 |
+
Quantization: 4-bit precision using BitsAndBytes (bnb) - Optimizing for accessibility without sacrificing logical reasoning accuracy.
|
155 |
+
Extended Reasoning Window: Supports inputs up to 16k tokens, crucial for accommodating the detailed context required for complex logical deductions.
|
156 |
+
|
157 |
+
Training Methodology: Honing Logical Acumen
|
158 |
+
|
159 |
+
Frameworks: Huggingface Transformers + TRL + Unsloth - A powerful combination for efficient training and reinforcement learning.
|
160 |
+
Data Sources: A meticulously curated collection of datasets specifically designed to challenge and refine logical reasoning skills, encompassing academic, legal, and formal logic domains.
|
161 |
+
Optimization Strategies:
|
162 |
+
LoRA (Low-Rank Adaptation): Enabling parameter-efficient fine-tuning, focusing on adapting the model for superior logical inference.
|
163 |
+
Reinforcement Learning from Human Feedback (RLHF): Guiding the model towards generating more logically sound and human-aligned reasoning steps.
|
164 |
+
|
165 |
+
Sphinx's Reasoning Toolkit: Capabilities in Action
|
166 |
+
|
167 |
+
Masterful Long-CoT Generation: Deconstructs and conquers multi-layered problems by constructing detailed, logically interconnected reasoning sequences.
|
168 |
+
Explanatory Power through Logic: Provides clear, step-by-step logical derivations for its outputs, enhancing trust and understanding.
|
169 |
+
Adaptable Logical Framework: Easily tailored to specialized reasoning tasks through targeted fine-tuning, enabling application in diverse logical domains.
|
170 |
+
|
171 |
+
Unlocking Potential: Applications Driven by Logic
|
172 |
+
|
173 |
+
Advanced Academic Research: Generating in-depth, logically structured analyses for complex scientific and philosophical inquiries.
|
174 |
+
Robust Legal Reasoning Assistance: Constructing and articulating multi-step legal arguments with precision and logical rigor.
|
175 |
+
Transformative STEM Education: Guiding learners through intricate mathematical and logical problems with clear, step-by-step explanations.
|
176 |
+
Transparent Cognitive AI Systems: Powering AI systems where explainability and logical justification are paramount for decision-making.# Open LLM Leaderboard Evaluation Results Detailed results can be found here! Summarized results can be found here!
|
177 |
+
|
178 |
+
---
|
179 |
## Use with llama.cpp
|
180 |
Install llama.cpp through brew (works on Mac and Linux)
|
181 |
|