Initial commit

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -1,3 +1,32 @@
 ---
+library_name: peft
 license: apache-2.0
 ---
+# minihf_evaluator_openllama_7b
+
+`minihf_evaluator_openllama_7b` is a LoRA instruct fine-tune of [OpenLLaMA 7B](https://huggingface.co/openlm-research/open_llama_7b).
+
+The sequence `<|end|>` was used to separate the prompt and response. The correct way to prompt the model is: `Does 2 + 2 = 4?<|end|>`. The tokenizer will prepend a BOS token (`<s>`) by default. The response will end with an EOS token (`</s>`).
+
+## Training procedure
+
+`minihf_evaluator_openllama_7b` was fine-tuned for 100,000 examples on 90% [Muennighoff/flan](https://huggingface.co/datasets/Muennighoff/flan) / 10% [databricks/databricks-dolly-15k](https://huggingface.co/datasets/databricks/databricks-dolly-15k) using batch size 4 per GPU on 8 40GB A100 GPUs. Examples where the prompt and response would not fit into 2,048 tokens were dropped. The fine-tuning was done using the following command:
+
+```bash
+accelerate launch make_evaluator.py --output-dir minihf_evaluator_openllama_7b
+```
+
+The following `bitsandbytes` quantization config was used during training:
+- load_in_8bit: False
+- load_in_4bit: True
+- llm_int8_threshold: 6.0
+- llm_int8_skip_modules: None
+- llm_int8_enable_fp32_cpu_offload: False
+- llm_int8_has_fp16_weight: False
+- bnb_4bit_quant_type: nf4
+- bnb_4bit_use_double_quant: True
+- bnb_4bit_compute_dtype: bfloat16
+
+### Framework versions
+
+- PEFT 0.4.0.dev0

adapter_config.json

@@ -0,0 +1,26 @@
+{
+  "base_model_name_or_path": "openlm-research/open_llama_7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 8,
+  "lora_dropout": 0.0,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 32,
+  "revision": null,
+  "target_modules": [
+    "self_attn.q_proj",
+    "self_attn.k_proj",
+    "self_attn.v_proj",
+    "self_attn.o_proj",
+    "mlp.gate_proj",
+    "mlp.up_proj",
+    "mlp.down_proj",
+    "lm_head"
+  ],
+  "task_type": null
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a4a1be941defab8d86563b05e463faca7e1b21b8a81659649dedbe24ad780c4
+size 324496576

make_evaluator.py

@@ -0,0 +1,255 @@
+#!/usr/bin/env python3
+
+"""Train a MiniHF evaluator model (instruction tuned LoRA)."""
+
+import argparse
+from functools import partial
+import os
+from pathlib import Path
+import sys
+
+os.environ["BITSANDBYTES_NOWELCOME"] = "1"
+
+import accelerate
+import datasets
+import datasets.distributed
+import peft
+import torch
+from torch import optim
+from torch.nn import functional as F
+from torch.utils import data
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+from tqdm import tqdm
+
+print = tqdm.external_write_mode()(print)
+
+
+def batch_to_tensors(batch, device="cpu"):
+    batch = [item["input_ids"] for item in batch]
+    seq_len = max(len(x) for x in batch)
+    input_ids = torch.zeros(len(batch), seq_len, dtype=torch.long, device=device)
+    attention_mask = torch.zeros(len(batch), seq_len, dtype=torch.long, device=device)
+    for i, x in enumerate(batch):
+        input_ids[i, : len(x)] = torch.tensor(x, dtype=torch.long, device=device)
+        attention_mask[i, : len(x)] = 1
+    return input_ids, attention_mask
+
+
+def weighted_mean(x, w=None, dim=None, keepdim=False, dtype=None):
+    w = x.new_tensor(1.0) if w is None else w
+    w = w.expand_as(x)
+    dim = tuple(range(x.ndim)) if dim is None else dim
+    num = torch.sum(x * w, dim=dim, keepdim=keepdim, dtype=dtype)
+    denom = torch.sum(w, dim=dim, keepdim=keepdim, dtype=dtype)
+    return num / denom
+
+
+class EndlessHFDataset(data.IterableDataset):
+    def __init__(self, dataset):
+        super().__init__()
+        self.dataset = dataset
+
+    def __iter__(self):
+        while True:
+            yield from self.dataset
+            self.dataset.set_epoch(self.dataset._epoch + 1)
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description=__doc__, formatter_class=argparse.ArgumentDefaultsHelpFormatter
+    )
+    parser.add_argument("--batch-size", type=int, default=4, help="batch size per process")
+    parser.add_argument("--examples", type=int, default=100000, help="train for n examples")
+    parser.add_argument("--output-dir", type=Path, default="evaluator", help="output directory")
+    parser.add_argument("--save-every", type=int, default=10000, help="save every n examples")
+    args = parser.parse_args()
+
+    dataset_seed = 100
+    lora_rank = 32
+    lr = 1e-4
+    max_len = 2048
+    model_name = "openlm-research/open_llama_7b"
+
+    # Initialize Accelerate
+    accelerator = accelerate.Accelerator(mixed_precision="bf16", dispatch_batches=False)
+    device = accelerator.device
+    print0 = accelerator.on_local_main_process(print)
+
+    # Load tokenizer
+    print0(f"### Loading tokenizer: {model_name}", file=sys.stderr)
+    tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=True)
+    tokenizer.pad_token = tokenizer.eos_token
+
+    # Load model
+    print0(f"### Loading model: {model_name}", file=sys.stderr)
+    bnb_config = BitsAndBytesConfig(
+        load_in_4bit=True,
+        bnb_4bit_compute_dtype=torch.bfloat16,
+        bnb_4bit_quant_type="nf4",
+        bnb_4bit_use_double_quant=True,
+    )
+    with accelerator.main_process_first():
+        model = AutoModelForCausalLM.from_pretrained(
+            model_name,
+            device_map="auto" if accelerator.num_processes == 1 else {"": device},
+            quantization_config=bnb_config,
+            torch_dtype=torch.bfloat16,
+            trust_remote_code=True,
+        )
+    accelerator.wait_for_everyone()
+
+    # Set up the LoRA
+    print0("### Setting up the LoRA", file=sys.stderr)
+    peft_config = peft.LoraConfig(
+        peft.TaskType.CAUSAL_LM,
+        inference_mode=False,
+        r=lora_rank,
+        lora_alpha=8,
+        lora_dropout=0.0,
+        target_modules=[
+            "self_attn.q_proj",
+            "self_attn.k_proj",
+            "self_attn.v_proj",
+            "self_attn.o_proj",
+            "mlp.gate_proj",
+            "mlp.up_proj",
+            "mlp.down_proj",
+            "lm_head",
+        ],
+    )
+    model = peft.get_peft_model(model, peft_config)
+    accelerator.wait_for_everyone()
+
+    # Set up the model
+    model.train()
+    model.gradient_checkpointing_enable()
+    model.enable_input_require_grads()
+    if accelerator.is_local_main_process:
+        model.print_trainable_parameters()
+
+    # Dataset helper functions
+    def combine_flan(row):
+        return row["inputs"] + "<|end|>" + row["targets"] + tokenizer.eos_token
+
+    def combine_dolly(row):
+        return (
+            row["context"]
+            + "\n\n"
+            + row["instruction"]
+            + "<|end|>"
+            + row["response"]
+            + tokenizer.eos_token
+        )
+
+    def to_tokens(combine_fn, row):
+        return tokenizer(combine_fn(row))
+
+    def exclude_too_long(row):
+        return len(row["input_ids"]) <= max_len
+
+    # Load dataset
+    print0("### Loading datasets", file=sys.stderr)
+    with accelerator.main_process_first():
+        dataset_1 = datasets.load_dataset("Muennighoff/flan", streaming=True)
+        dataset_2 = datasets.load_dataset("databricks/databricks-dolly-15k", streaming=True)
+    accelerator.wait_for_everyone()
+    dataset_1 = dataset_1["train"].map(partial(to_tokens, combine_flan))
+    dataset_2 = dataset_2["train"].map(partial(to_tokens, combine_dolly))
+    dataset = (
+        datasets.interleave_datasets([dataset_1, dataset_2], probabilities=[0.9, 0.1])
+        .filter(exclude_too_long)
+        .shuffle(seed=dataset_seed)
+        .select_columns(["input_ids"])
+    )
+    dataset = datasets.distributed.split_dataset_by_node(
+        dataset, accelerator.process_index, accelerator.num_processes
+    )
+    dataloader = data.DataLoader(
+        EndlessHFDataset(dataset),
+        batch_size=args.batch_size,
+        collate_fn=batch_to_tensors,
+        drop_last=True,
+    )
+
+    # Set up optimizer
+    opt = optim.Adam(model.parameters(), lr=lr, betas=(0.9, 0.99))
+
+    # Wrap objects
+    model, opt, dataloader = accelerator.prepare(model, opt, dataloader)
+
+    # Test max sequence length
+    print0("### Testing max sequence length", file=sys.stderr)
+    input_ids = torch.zeros([args.batch_size, max_len], dtype=torch.long, device=device)
+    attention_mask = torch.ones([args.batch_size, max_len], dtype=torch.long, device=device)
+    outputs = model(input_ids, attention_mask=attention_mask, use_cache=False)
+    accelerator.backward(outputs.logits.sum() * 0)
+    opt.zero_grad()
+    torch.cuda.empty_cache()
+
+    def save_model():
+        print0("### Saving model", file=sys.stderr)
+        accelerator.wait_for_everyone()
+        if accelerator.is_main_process:
+            unwrapped_model = accelerator.unwrap_model(model)
+            unwrapped_model.save_pretrained(args.output_dir, safe_serialization=True)
+            tokenizer.save_pretrained(args.output_dir)
+
+    # Train
+    print0("### Training", file=sys.stderr)
+    examples = 0
+    last_save = 0
+    pbar = tqdm(
+        disable=not accelerator.is_local_main_process,
+        total=args.examples,
+        unit="ex",
+        smoothing=0.01,
+    )
+
+    try:
+        for batch in dataloader:
+            input_ids, attention_mask = batch
+            with accelerator.accumulate(model):
+                # Forward pass
+                outputs = model(
+                    input_ids[:, :-1],
+                    attention_mask=attention_mask[:, :-1],
+                    use_cache=False,
+                )
+                losses = F.cross_entropy(
+                    outputs.logits.transpose(-1, -2),
+                    input_ids[:, 1:],
+                    reduction="none",
+                )
+                mask = attention_mask[:, :-1] * attention_mask[:, 1:]
+                loss = weighted_mean(losses, mask, dtype=torch.float32)
+
+                # Backward pass and optimizer step
+                accelerator.backward(loss)
+                opt.step()
+                opt.zero_grad()
+
+            global_batch_size = args.batch_size * accelerator.num_processes
+            examples += global_batch_size
+            pbar.update(global_batch_size)
+
+            global_loss = accelerator.reduce(loss, "mean")
+            print0(f"examples: {examples}, loss: {global_loss.item():g}")
+
+            if examples >= args.examples:
+                save_model()
+                break
+
+            if examples - last_save >= args.save_every:
+                save_model()
+                last_save += args.save_every
+
+    except KeyboardInterrupt:
+        pass
+
+    finally:
+        pbar.close()
+
+
+if __name__ == "__main__":
+    main()

special_tokens_map.json

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

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab1b681ec7fc02fed5edd3026687d7a692a918c4dd8e150ca2e3994a6229843b
+size 534194

tokenizer_config.json

@@ -0,0 +1,33 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "model_max_length": 2048,
+  "pad_token": null,
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}