Update README.md

README.md

@@ -3,15 +3,15 @@ pipeline_tag: text-generation
 inference: false
 license: apache-2.0
 datasets:
-- codeparrot/github-code-clean
-- bigcode/starcoderdata
-# - Stackexchange
-# - CommonCrawl
-- open-web-math/open-web-math
-- math-ai/StackMathQA
-# - Arxiv
-# - Wikipedia
-# - conceptofmind/FLAN_2022 # Original link is broken, we used IBM's filtered version
+  - bigcode/commitpackft
+  - TIGER-Lab/MathInstruct
+  - meta-math/MetaMathQA
+  - glaiveai/glaive-code-assistant-v3
+  - glaive-function-calling-v2
+  - bugdaryan/sql-create-context-instruction
+  - garage-bAInd/Open-Platypus
+  - nvidia/HelpSteer
+  - bigcode/self-oss-instruct-sc2-exec-filter-50k
 metrics:
 - code_eval
 library_name: transformers
@@ -19,7 +19,7 @@ tags:
 - code
 - granite
 model-index:
-- name: granite-8b-code-base-128k
+- name: granite-8B-Code-instruct-128k
   results:
   - task:
       type: text-generation
@@ -29,7 +29,7 @@ model-index:
     metrics:
     - name: pass@1
       type: pass@1
-      value: 43.1
+      value: 62.2
       verified: false
   - task:
       type: text-generation
@@ -39,7 +39,7 @@ model-index:
     metrics:
     - name: pass@1
       type: pass@1
-      value: 40.2
+      value: 51.4
       verified: false
   - task:
       type: text-generation
@@ -49,7 +49,7 @@ model-index:
     metrics:
     - name: pass@1
       type: pass@1
-      value: 28.2
+      value: 38.9
       verified: false
   - task:
       type: text-generation
@@ -59,7 +59,7 @@ model-index:
     metrics:
     - name: pass@1
       type: pass@1
-      value: 25.2
+      value: 38.3
       verified: false
   - task:
       type: text-generation
@@ -69,7 +69,7 @@ model-index:
     metrics:
     - name: pass@1 (thresh=0.5)
       type: pass@1 (thresh=0.5)
-      value: 65.0
+      value: 73.0
       verified: false
   - task:
       type: text-generation
@@ -79,7 +79,7 @@ model-index:
     metrics:
     - name: pass@1 (thresh=0.5)
       type: pass@1 (thresh=0.5)
-      value: 35.0
+      value: 37.0
       verified: false
   - task:
       type: text-generation
@@ -89,7 +89,7 @@ model-index:
     metrics:
     - name: pass@1 (thresh=0.5)
       type: pass@1 (thresh=0.5)
-      value: 39.0
+      value: 73.0
       verified: false
   - task:
       type: text-generation
@@ -99,7 +99,7 @@ model-index:
     metrics:
     - name: pass@1 (thresh=0.5)
       type: pass@1 (thresh=0.5)
-      value: 40.0
+      value: 62.0
       verified: false
   - task:
       type: text-generation
@@ -109,97 +109,17 @@ model-index:
     metrics:
     - name: pass@1 (thresh=0.5)
       type: pass@1 (thresh=0.5)
-      value: 54.0
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: lcc  
-        name:  LCC (Balanced)
-    metrics:
-    - name: Exact Match@4K 
-      type: Exact Match@4K
-      value: 56.5
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: lcc  
-        name:  LCC (Balanced)
-    metrics:
-    - name: Exact Match@8K 
-      type: Exact Match@8K
-      value: 60.1
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: lcc  
-        name:  LCC (Balanced)
-    metrics:
-    - name: Exact Match@16K 
-      type: Exact Match@16K
-      value: 51.8
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: lcc  
-        name:  LCC (Balanced)
-    metrics:
-    - name: Exact Match@32K 
-      type: Exact Match@32K
-      value: 57.4
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: repobench  
-        name: RepoBench-P (Balanced)
-    metrics:
-    - name: Exact Match@4K 
-      type: Exact Match@4K
-      value: 42.7
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: repobench  
-        name: RepoBench-P (Balanced)
-    metrics:
-    - name: Exact Match@8K 
-      type: Exact Match@8K
-      value: 44.0
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: repobench  
-        name: RepoBench-P (Balanced)
-    metrics:
-    - name: Exact Match@16K 
-      type: Exact Match@16K
-      value: 44.8
-      verified: false
-  - task:
-      type: text-generation
-    dataset:
-        type: repobench  
-        name: RepoBench-Pn(Balanced)
-    metrics:
-    - name: Exact Match@32K 
-      type: Exact Match@32K
-      value: 44.5
+      value: 63.0
       verified: false
 ---
 
+
 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/62cd5057674cdb524450093d/1hzxoPwqkBJXshKVVe6_9.png)
 
-# Granite-8B-Code-Base-128K
+# Granite-8B-Code-Instruct-128K
 
 ## Model Summary
-**Granite-8B-Code-Base-128K** extends the context length of Granite-8B-Code-Base from 4K to 128K with continual pretraining using the original training data but with repository-level file packing and per-language length upsampling, that we found to be critical for long-context pretraining. 
-We adopt an progressive training strategy where we doubled the context window until it reached the desired length of 128K by appropriately adjusting RoPE theta. We trained on 4B tokens total for all stages, which is only 0.1% of Granite-8B-Code-Base's original pre-training data. 
+**Granite-8B-Code-Instruct-128K** is a 3B parameter long-context instruct model fine tuned from *Granite-8B-Code-Base-128K* on a combination of **permissively licensed** data used in training the original Granite code instruct models, in addition to synthetically generated code instruction datasets tailored for solving long context problems. By exposing the model to both short and long context data, we aim to enhance its long-context capability without sacrificing code generation performance at short input context.
 
 - **Developers:** IBM Research
 - **GitHub Repository:** [ibm-granite/granite-code-models](https://github.com/ibm-granite/granite-code-models)
@@ -209,29 +129,34 @@ We adopt an progressive training strategy where we doubled the context window un
 
 ## Usage
 ### Intended use
-Prominent enterprise use cases of LLMs in software engineering productivity with 128K context length support that includes code generation, code explanation, code fixing, generating unit tests, generating documentation, addressing technical debt issues, vulnerability detection, code translation, and more. All Granite Code Base models, including the **3B parameter model**, are able to handle these tasks as they were trained on a large amount of code data from 116 programming languages. 
+The model is designed to respond to coding related instructions over long-conext input and can be used to build coding assistants.
+
+<!-- TO DO: Check starcoder2 instruct code example that includes the template https://huggingface.co/bigcode/starcoder2-15b-instruct-v0.1 -->
 
 ### Generation
-This is a simple example of how to use **Granite-8B-Code-Base-128K** model.
+This is a simple example of how to use **Granite-8B-Code-Instruct** model.
 
 ```python
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 device = "cuda" # or "cpu"
-model_path = "ibm-granite/granite-8b-code-base-128K"
+model_path = "ibm-granite/granite-8B-Code-instruct-128k"
 tokenizer = AutoTokenizer.from_pretrained(model_path)
 # drop device_map if running on CPU
 model = AutoModelForCausalLM.from_pretrained(model_path, device_map=device)
 model.eval()
 # change input text as desired
-input_text = "def generate():"
+chat = [
+    { "role": "user", "content": "Write a code to find the maximum value in a list of numbers." },
+]
+chat = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
 # tokenize the text
-input_tokens = tokenizer(input_text, return_tensors="pt")
+input_tokens = tokenizer(chat, return_tensors="pt")
 # transfer tokenized inputs to the device
 for i in input_tokens:
     input_tokens[i] = input_tokens[i].to(device)
 # generate output tokens
-output = model.generate(**input_tokens)
+output = model.generate(**input_tokens, max_new_tokens=100)
 # decode output tokens into text
 output = tokenizer.batch_decode(output)
 # loop over the batch to print, in this example the batch size is 1
@@ -239,11 +164,14 @@ for i in output:
     print(i)
 ```
 
+<!-- TO DO: Check this part -->
 ## Training Data
-Starting from the base Granite model, this model was further pretrained on repository-level code data with per-language oversampling, allowing it to effectively utilize up to 128K tokens of context. This continued training stage focused on a curated selection of programming languages, such as Python, C, C++, Go, Java, JavaScript, and TypeScript.
-
+Granite Code Instruct models are trained on a mix of short and long context data as follows.
+* Short-Context Instruction Data: [CommitPackFT](https://huggingface.co/datasets/bigcode/commitpackft), [BigCode-SC2-Instruct](bigcode/self-oss-instruct-sc2-exec-filter-50k), [MathInstruct](https://huggingface.co/datasets/TIGER-Lab/MathInstruct), [MetaMathQA](https://huggingface.co/datasets/meta-math/MetaMathQA), [Glaive-Code-Assistant-v3](https://huggingface.co/datasets/glaiveai/glaive-code-assistant-v3), [Glaive-Function-Calling-v2](https://huggingface.co/datasets/glaiveai/glaive-function-calling-v2), [NL2SQL11](https://huggingface.co/datasets/bugdaryan/sql-create-context-instruction), [HelpSteer](https://huggingface.co/datasets/nvidia/HelpSteer), [OpenPlatypus](https://huggingface.co/datasets/garage-bAInd/Open-Platypus) including a synthetically generated dataset for API calling and multi-turn code interaction with execution feedback. We also include a collection of hardcoded prompts to ensure our model generates correct outputs given inquiries about its name or developers.
+* Long-Context Instruction Data: A synthetically-generated dataset by bootstrapping the repository-level file-packed documents through Granite-8b-Code-Instruct to improve long-context capability of the model.
+  
 ## Infrastructure
 We train the Granite Code models using two of IBM's super computing clusters, namely Vela and Blue Vela, both outfitted with NVIDIA A100 and H100 GPUs respectively. These clusters provide a scalable and efficient infrastructure for training our models over thousands of GPUs.
 
 ## Ethical Considerations and Limitations
-The use of Large Language Models involves risks and ethical considerations people must be aware of. Regarding code generation, caution is urged against complete reliance on specific code models for crucial decisions or impactful information as the generated code is not guaranteed to work as intended. **Granite-8B-Code-Base-128K** model is not the exception in this regard. Even though this model is suited for multiple code-related tasks, it has not undergone any safety alignment, there it may produce problematic outputs. Additionally, it remains uncertain whether smaller models might exhibit increased susceptibility to hallucination in generation scenarios by copying source code verbatim from the training dataset due to their reduced sizes and memorization capacities. This aspect is currently an active area of research, and we anticipate more rigorous exploration, comprehension, and mitigations in this domain. Regarding ethics, a latent risk associated with all Large Language Models is their malicious utilization. We urge the community to use **Granite-8B-Code-Base-128K** model with ethical intentions and in a responsible way. 
+Granite code instruct models are primarily finetuned using instruction-response pairs across a specific set of programming languages. Thus, their performance may be limited with out-of-domain programming languages. In this situation, it is beneficial providing few-shot examples to steer the model's output. Moreover, developers should perform safety testing and target-specific tuning before deploying these models on critical applications. The model also inherits ethical considerations and limitations from its base model. For more information, please refer to *[Granite-8B-Code-Base-128K](https://huggingface.co/ibm-granite/granite-8B-Code-base-128k)* model card.