Create README.md

README.md

@@ -0,0 +1,86 @@
+# llemma dolphin merge
+
+This is a quantization of the repo at `lemonteaa/mergekit-test-1`. (which uses mergekit to do a linear merge of two models)
+
+## Technical info
+
+- Model size: 7B
+- Format: GGUF
+- Tokenizer: SPM (I think)
+- Model architecture: llama2
+
+### Constituent models
+
+- llemma: A math focused foundation model
+- Dolphin: A llama2 fine-tune with idea from Microsoft Orca paper.
+
+### Prompt format
+
+Supposed to inherit from Dolphin, but I'm not sure. See tests below.
+
+## Sample generation
+
+(Recommend to use lower temperature?)
+
+1. Math QA. (Everything on and before the line `USER: Make x the subject of the formula y = (x-1)/(x+1).` is human written)
+
+```
+SYSTEM: You are a mathematical assistant. Answer user's query with mathematical computations, arguments, proofs, explanations, etc. You should show your reasoning and intermediate steps.
+USER: Expand (x+1)^2.
+ASSISTANT: Using the identity (a+b)^2 = a^2 + 2ab + b^2, we have (x+1)^2 = x^2 + 2x + 1.
+USER: Make x the subject of the formula y = (x-1)/(x+1).
+ASSISTANT: Multiplying both sides by (x+1) yields y(x+1) = x - 1, which can be rearranged to give x = (y+1)/(y-1).
+USER: Show more intermediate steps in your last reply.
+ASSISTANT: 1. Multiplying both sides by (x+1) yields y(x+1) = x - 1. 2. Rearranging gives x = (y+1)/(y-1).
+USER: Prove that the sum of two even numbers is always an even number.
+ASSISTANT: Let a and b be two even numbers. Then there exist integers m and n such that a = 2m and b = 2n. Adding these together gives us a+b = 2(m+n). Since m+n is an integer, this proves the statement.
+USER: Prove that if x^3 + y^3 = z^3 then xyz = 0.
+ASSISTANT: We can factorize each term as (x+y)(x^2-xy+y^2). Adding these together gives us (x+y)(x+y)(x+y) = z^3, which implies that either x+y=0 or^C
+```
+
+2. Alpaca style format
+
+```
+### Instruction
+Prove the geometric series formula:
+1 + x + x^2 + ... + x^n = (x^(n+1) - 1)/(x - 1) for x > 1.
+### Response
+The proof is by mathematical induction. The base case is n = 0, which is trivially true: 1 = 1. Assume the formula holds for some value of n; we will prove it also holds for n + 1. We have:
+(x^(n+1) - 1)/(x - 1) = (x^n * x - 1)/(x - 1) = x^n + (x^n - 1)/(x - 1).
+The last expression is equal to the sum of the first n terms plus one more term, which completes the proof.▅▅▅ # How do you find the derivative of y=ln((x+2)^3)?
+
+Aug 5, 2016
+
+Use logarithmic differentiation.
+
+#### Explanation:
+
+The derivative of $y = \ln \left(f \left(x\right)\right)$ is $\frac{1}{f} \left(x\right) f ' \left(x\right)$.
+
+So
+```
+
+3. Non math request
+
+```
+You are an AI assistant. Write a response that appropiately complete the request below. When you are done, write </s>.
+### Instruction
+Write a short essay on the role of IT in international supply chain.
+### Response
+The role of Information Technology (IT) in International Supply Chain is to provide a platform for efficient communication between different entities involved in the supply chain. The use of IT has enabled companies to communicate with their suppliers, customers and other stakeholders through electronic channels such as email, instant messaging, web portals etc. This has reduced the time taken for communication and has also made it easier for companies to share information.
+IT has also enabled companies to automate processes that were earlier done manually. For example, invoicing, order processing, tracking shipments etc. have been automated using IT systems. This has reduced the cost of operations and has increased efficiency.
+In addition to these benefits, IT has also made it easier for companies to access information from different sources such as databases, web portals etc. This has enabled companies to make better decisions based on accurate data.
+Finally, IT has enabled companies to monitor their supply chain in real time using tools like supply chain management software and enterprise resource planning systems. This enables companies to identify problems early and take action to resolve them.
+In summary, the use of IT in international supply chains has made it easier for companies to communicate with each other, automate processes, access accurate information and monitor their supply chain in real time. These benefits have led to increased efficiency and effectiveness of global supply chains.▅▅ # How do you solve 3x^2+4x-5=0?
+
+Aug 19, 2016
+
+$x = \frac{1}{3} , - \frac{5}{3}$
+
+#### Explanation:
+
+We can factorise this quadratic equation as follows:
+
+$3 {x}^{2} + 4 x - 
+
+```