Upload 8 files

README.md

@@ -1,3 +1,189 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+library_name: peft
+tags:
+- alignment-handbook
+- trl
+- sft
+- generated_from_trainer
+base_model: mistralai/Mistral-7B-v0.1
+model-index:
+- name: Cimphony-Mistral-Law-7B
+  results:
+  - task: 
+      type: text-generation
+    dataset: 
+      type: cais/mmlu
+      name: MMLU
+    metrics:
+    - name: International Law
+      type: accuracy
+      value: 0.802
+      verified: false
+  - task: 
+      type: text-generation
+    dataset: 
+      type: cais/mmlu
+      name: MMLU
+    metrics:
+    - name: Jurisprudence
+      type: accuracy
+      value: 0.704
+      verified: false
+  - task: 
+      type: text-generation
+    dataset: 
+      type: cais/mmlu
+      name: MMLU
+    metrics:
+    - name: Professional Law
+      type: accuracy
+      value: 0.416
+      verified: false
+  - task: 
+      type: text-generation
+    dataset: 
+      type: coastalcph/lex_glue
+      name: LexGLUE
+    metrics:
+    - name: ECtHR A
+      type: balanced accuracy
+      value: 0.631
+      verified: false
+  - task: 
+      type: text-generation
+    dataset: 
+      type: coastalcph/lex_glue
+      name: LexGLUE
+    metrics:
+    - name: LEDGAR
+      type: balanced accuracy
+      value: 0.741
+      verified: false
+  - task: 
+      type: text-generation
+    dataset: 
+      type: coastalcph/lex_glue
+      name: LexGLUE
+    metrics:
+    - name: CaseHOLD
+      type: accuracy
+      value: 0.776
+      verified: false
+  - task: 
+      type: text-generation
+    dataset: 
+      type: coastalcph/lex_glue
+      name: LexGLUE
+    metrics:
+    - name: Unfair-ToS
+      type: balanced accuracy
+      value: 0.809
+      verified: false
+      
+pipeline_tag: text-generation
+---
+
+# Cimphony-Mistral-Law-7B
+
+We introduce Cimphony-Mistral-Law-7B, a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1).
+
+Cimphony’s LLMs present state-of-the-art performance on legal benchmarks, suppressing models trained on a much larger corpus with significantly more resources, and in some cases even GPT-4, OpenAI’s flagship model.
+
+![image/png](https://assets-global.website-files.com/64bde220a9b6312524909e4f/654a74e810c7aa04e73056c6_PNG-Final%20File-Logo-Cimphony-Vektora-Horizontal-01-p-500.png)
+
+## Model description
+
+The model was trained on 600M tokens. We use novel methods to expose the model to this corpus during training, blending a variety of legal reading comprehension tasks, as well as general language data.
+
+
+## Legal Evaluation Results
+
+We evaluate on the legal splits of the MMLU benchmark, as well as LexGLUE. While both are multiple option benchmarks, prompts were adapted so that the models output a single answer. In some cases, additional post-processing was required.
+
+Benchmarks for which the labels were A-E multiple-choice options use an accuracy mertic. Benchmarks that have a closed list of options (e.g. Unfair-ToS) use a balanced-accuracy metric, as classes may not be balanced.
+
+| Model / Benchmark                  | International Law (MMLU)  | Jurisprudence (MMLU)  | Professional law (MMLU)  | ECtHR A (LexGlue)  | LEDGAR (LexGlue)  | CaseHOLD (LexGlue)  | Unfair-ToS (LexGlue)   |
+|:-----------------------------------|:--------------------------|:----------------------|:-------------------------|:-------------------|:------------------|:--------------------|:-----------------------|
+| Mistral-7B-Instruct-v0.2           | 73.6%                     | 69.4%                 | 41.2%                    |  67.5%             | 50.6%             |  56.3%              | 36.6%                  |
+| AdaptLLM                           | 57.0%                     | 52.8%                 | 36.1%                    | 51.9%              | 46.3%             |  50.0%              | 51.3%                  |
+| Saul-7B                            | 69.4%                     | 63.0%                 | **43.2%**                | **71.2%**          | 55.9%             | 65.8%               | 80.3%                  |
+|<tr style="background-color:yellow;"><td>Cimphony-7B</td><td>**80.2%**</td><td>**70.4%**</td><td>41.6%</td><td>63.1%</td><td>**74.1%**</td><td>**77.6%**</td><td>**80.9%**</td></tr>|
+ 
+## Training and evaluation data
+
+Following the framework presented in [AdaptLLM](https://huggingface.co/AdaptLLM/law-chat), we convert the raw legal text into reading comprehension. Taking inspiration from human learning via reading comprehension - practice after reading improves the ability to answer questions based on the learned knowledge.
+
+We developed a high-quality prompt database, considering the capabilities we’d like the model to possess. LLMs were prompt with the raw text and a collection of prompts, and it returned answers, additional questions, and transformations relevant to the input data. With further post-processing of these outputs, we created our legal reading comprehension dataset.
+
+
+| Domain             | Dataset             | Tokens | License     |
+|:-------------------|:--------------------|:------:|:------------|
+| Legal              | The Pile (FreeLaw)  | 180M   | MIT         |
+| Legal              | LexGlue             | 108M   | CC-BY-4.0   |
+| Legal              | USClassActions      | 12M    | GPL-3.0     |
+| Math (CoT)         | AQUA-RAT            | 3M     | Apache-2.0  |
+| Commonsense (CoT)  | ECQA                | 2.4M   | Apache-2.0  |
+| Reasoning (CoT)    | EntailmentBank      | 1.8M   | Apache-2.0  |
+| Chat               | UltraChat           | 90M    | MIT         |
+| Code               | Code-Feedback       | 36M    | Apache-2.0  |
+| Instruction        | OpenOrca            | 180M   | MIT         |
+
+
+## Intended uses & limitations
+
+This model can be used for use cases involving legal domain text generation.
+
+As with any language model, users must not solely relay on model generations. This model has not gone through a human-feedback alignment (RLHF). The model may generate responses containing hallucinations and biases.
+
+Example use:
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+
+tokenizer = AutoTokenizer.from_pretrained("iarbel/mistral-law-7b-beta")
+model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1")
+model = PeftModel.from_pretrained(model, "iarbel/mistral-law-7b-beta")
+
+# Put your input here:
+user_input = '''Question: What can you tell me about ex post facto laws?'''
+
+# Apply the prompt template
+prompt = tokenizer.apply_chat_template(user_input, tokenize=False)
+
+inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False).input_ids.to(model.device)
+outputs = model.generate(input_ids=inputs, max_length=4096)[0]
+
+answer_start = int(inputs.shape[-1])
+pred = tokenizer.decode(outputs[answer_start:], skip_special_tokens=True)
+
+print(f'### User Input:\n{user_input}\n\n### Assistant Output:\n{pred}')
+```
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 0.0005
+- train_batch_size: 8
+- eval_batch_size: 24
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 4
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 128
+- total_eval_batch_size: 96
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.05
+- num_epochs: 1
+
+
+### Framework versions
+
+- PEFT 0.8.2
+- Transformers 4.37.2
+- Pytorch 2.1.2+cu121
+- Datasets 2.14.6
+- Tokenizers 0.15.2

adapter_config.json

@@ -0,0 +1,31 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-v0.1",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "o_proj",
+    "q_proj",
+    "down_proj",
+    "k_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74ab9437033b622089e58b4cb3b83e59a4c7e3ca6693272696ba93923be598a8
+size 260098992

config.json

@@ -0,0 +1,26 @@
+{
+  "_name_or_path": "mistralai/Mistral-7B-v0.1",
+  "architectures": [
+    "MistralForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 32768,
+  "model_type": "mistral",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-05,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.37.2",
+  "use_cache": true,
+  "vocab_size": 32000
+}

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<unk>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
+size 493443

tokenizer_config.json

@@ -0,0 +1,42 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 4096,
+  "pad_token": "<unk>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}