diff --git a/src/app/(docs)/_guides/contribute/overview/page.mdx b/src/app/(docs)/_guides/contribute/overview/page.mdx index 5a5f6001..bdc3359f 100644 --- a/src/app/(docs)/_guides/contribute/overview/page.mdx +++ b/src/app/(docs)/_guides/contribute/overview/page.mdx @@ -20,10 +20,10 @@ If you are interested in contributing to our [official docs](https://docs.mistra You can easily help by: -- fix a typo -- clarify a section -- document an underdocumented feature -- update a section that should have been updated +- fixing a typo +- clarifying a section +- documenting an underdocumented feature +- updating a section that should have been updated - ... ## Contributing to the code clients diff --git a/src/app/(docs)/capabilities/embeddings/rag_quickstart/page.mdx b/src/app/(docs)/capabilities/embeddings/rag_quickstart/page.mdx index f85439a2..fe01a6b9 100644 --- a/src/app/(docs)/capabilities/embeddings/rag_quickstart/page.mdx +++ b/src/app/(docs)/capabilities/embeddings/rag_quickstart/page.mdx @@ -24,7 +24,7 @@ In this guide, we will walk through a very basic example of RAG, you can find mo -This section aims to guide you through the process of building a basic RAG from scratch. We have two goals: firstly, to offer users a comprehensive understanding of the internal workings of RAG and demystify the underlying mechanisms; secondly, to empower you with the essential foundations needed to build an RAG using the minimum required dependencies. +This section aims to guide you through the process of building a basic RAG from scratch. We have two goals: firstly, to offer users a comprehensive understanding of the internal workings of RAG and demystify the underlying mechanisms; secondly, to empower you with the essential foundations needed to build a RAG using the minimum required dependencies. Import needed packages @@ -128,7 +128,7 @@ text_embeddings = np.array([get_text_embedding(chunk) for chunk in chunks]) Load into a vector database -Once we get the text embeddings, a common practice is to store them in a vector database for efficient processing and retrieval. There are several vector database to choose from. In our simple example, we are using an open-source vector database Faiss, which allows for efficient similarity search. +Once we get the text embeddings, a common practice is to store them in a vector database for efficient processing and retrieval. There are several vector databases to choose from. In our simple example, we are using an open-source vector database Faiss, which allows for efficient similarity search. With Faiss, we instantiate an instance of the Index class, which defines the indexing structure of the vector database. We then add the text embeddings to this indexing structure. @@ -186,7 +186,7 @@ retrieved_chunk = [chunks[i] for i in I.tolist()[0]] :::note - **Retrieval methods**: There are a lot different retrieval strategies. In our example, we are showing a simple similarity search with embeddings. Sometimes when there is metadata available for the data, it's better to filter the data based on the metadata first before performing similarity search. There are also other statistical retrieval methods like TF-IDF and BM25 that use frequency and distribution of terms in the document to identify relevant text chunks. -- **Retrieved document**: Do we always retrieve individual text chunk as it is? Not always. +- **Retrieved document**: Do we always retrieve an individual text chunk as is? Not always. - Sometime, we would like to include more context around the actual retrieved text chunk. We call the actual retrieved text chunk "child chunk" and our goal is to retrieve a larger "parent chunk" that the "child chunk" belongs to. - On occasion, we might also want to provide weights to our retrieve documents. For example, a time-weighted approach would help us retrieve the most recent document. - One common issue in the retrieval process is the "lost in the middle" problem where the information in the middle of a long context gets lost. Our models have tried to mitigate this issue. For example, in the passkey task, our models have demonstrated the ability to find a "needle in a haystack" by retrieving a randomly inserted passkey within a long prompt, up to 32k context length. However, it is worth considering experimenting with reordering the document to determine if placing the most relevant chunks at the beginning and end leads to improved results. diff --git a/src/app/(docs)/getting-started/models/overview/_page.mdx b/src/app/(docs)/getting-started/models/overview/_page.mdx index 13565b1b..205ec235 100644 --- a/src/app/(docs)/getting-started/models/overview/_page.mdx +++ b/src/app/(docs)/getting-started/models/overview/_page.mdx @@ -27,7 +27,7 @@ Mistral provides two types of models: open models and premier models. | Ministral 8B | :heavy_check_mark:
[Mistral Research License](https://mistral.ai/licenses/MRL-0.1.md)| :heavy_check_mark: |Powerful edge model with extremely high performance/price ratio. Learn more in our [blog post](https://mistral.ai/news/ministraux/) | 128k | `ministral-8b-2410` | 24.10| | Mistral Medium 3 | | :heavy_check_mark: | Our frontier-class multimodal model released May 2025. Learn more in our [blog post](https://mistral.ai/news/mistral-medium-3/) | 128k | `mistral-medium-2505` | 25.05| | Codestral 2501 | | :heavy_check_mark: | Our cutting-edge language model for coding with the second version released January 2025, Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation. Learn more in our [blog post](https://mistral.ai/news/codestral-2501/) | 256k | `codestral-2501` | 25.01| -| Mistral Large 2.1 |:heavy_check_mark:
[Mistral Research License](https://mistral.ai/licenses/MRL-0.1.md)| :heavy_check_mark: | Our top-tier large model for high-complexity tasks with the lastest version released November 2024. Learn more in our [blog post](https://mistral.ai/news/pixtral-large/) | 128k | `mistral-large-2411` | 24.11| +| Mistral Large 2.1 |:heavy_check_mark:
[Mistral Research License](https://mistral.ai/licenses/MRL-0.1.md)| :heavy_check_mark: | Our top-tier large model for high-complexity tasks with the latest version released November 2024. Learn more in our [blog post](https://mistral.ai/news/pixtral-large/) | 128k | `mistral-large-2411` | 24.11| | Pixtral Large |:heavy_check_mark:
[Mistral Research License](https://mistral.ai/licenses/MRL-0.1.md)| :heavy_check_mark: | Our first frontier-class multimodal model released November 2024. Learn more in our [blog post](https://mistral.ai/news/pixtral-large/) | 128k | `pixtral-large-2411` | 24.11| | Mistral Small 2| :heavy_check_mark:
[Mistral Research License](https://mistral.ai/licenses/MRL-0.1.md) | :heavy_check_mark: | Our updated small version, released September 2024. Learn more in our [blog post](https://mistral.ai/news/september-24-release) | 32k | `mistral-small-2407` | 24.07| | Mistral Embed | | :heavy_check_mark: | Our state-of-the-art semantic for extracting representation of text extracts | 8k | `mistral-embed` | 23.12| diff --git a/src/app/(docs)/getting-started/models/picking/_page.mdx b/src/app/(docs)/getting-started/models/picking/_page.mdx index 484e88ed..3dca0abd 100644 --- a/src/app/(docs)/getting-started/models/picking/_page.mdx +++ b/src/app/(docs)/getting-started/models/picking/_page.mdx @@ -29,7 +29,7 @@ As a general rule, if you are new to using Large Language Models, you can always ## Use cases ### Mistral Nemo: Simple tasks that one can do in bulk -Mistral Nemo is the ideal choice for simpe tasks that one can do in builk - like Classification, Customer Support, or Text Generation. It offers excellent performance at an affordable price point. For instance, it can be effectively used for a classification task to classify if an email is spam or not: +Mistral Nemo is the ideal choice for simple tasks that one can do in bulk - like Classification, Customer Support, or Text Generation. It offers excellent performance at an affordable price point. For instance, it can be effectively used for a classification task to classify if an email is spam or not: **Prompt:** ``` @@ -41,7 +41,7 @@ Classify the following email to determine if it is spam or not. Only respond wit All of our models can accurately classify this email correctly as “Spam”. Mistral Nemo is capable to provide the correct classification as the larger models. So it is the most efficient and affordable choice for this kind of tasks. ### Mistral Small: Intermediate tasks that require language transformation -Mistral Small is the ideal for intermediate tasks that require moderate reasoning - like Data extraction, Summarizing a Document, Writing a Job Description, or Writing Product Descriptions. Mistral Small strikes a balance between performance and capability, making it suitable for a wide range of tasks that only require language transformaion. For example, Mistral Small can write an email: +Mistral Small is the ideal for intermediate tasks that require moderate reasoning - like Data extraction, Summarizing a Document, Writing a Job Description, or Writing Product Descriptions. Mistral Small strikes a balance between performance and capability, making it suitable for a wide range of tasks that only require language transformation. For example, Mistral Small can write an email: **Prompt:** ``` @@ -138,7 +138,7 @@ It may seem counterintuitive because we often think of iron as being a heavy mat - **Coding** -Mistral Large, the top performer in coding tasks, is the ideal choice for users who priortize coding capabilities in their model selection. +Mistral Large, the top performer in coding tasks, is the ideal choice for users who prioritize coding capabilities in their model selection. Here is an example of Mistral Large giving the correct code following the prompt.