Text Generation
Transformers
Safetensors
English
llama
astheral
instruct
qlora
conversational
text-generation-inference
Instructions to use SEE7/Astheral with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use SEE7/Astheral with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="SEE7/Astheral") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("SEE7/Astheral") model = AutoModelForCausalLM.from_pretrained("SEE7/Astheral", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=256) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use SEE7/Astheral with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "SEE7/Astheral" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "SEE7/Astheral", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/SEE7/Astheral
- SGLang
How to use SEE7/Astheral with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "SEE7/Astheral" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "SEE7/Astheral", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "SEE7/Astheral" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "SEE7/Astheral", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use SEE7/Astheral with Docker Model Runner:
docker model run hf.co/SEE7/Astheral
Release tested Astheral V3: matched BF16 improvement and independent regression checks
Browse files- README.md +17 -45
- adapter/adapter_config.json +8 -8
- adapter/adapter_model.safetensors +2 -2
- artifact_manifest.json +99 -27
- config.json +1 -1
- evaluation.json +91 -143
- generation_config.json +3 -7
- holdout_eval.py +30 -9
- holdout_evaluation.json +11 -5
- holdout_outputs.json +22 -22
- model-00001-of-00005.safetensors +1 -1
- model-00002-of-00005.safetensors +1 -1
- model-00003-of-00005.safetensors +1 -1
- model-00004-of-00005.safetensors +1 -1
- paired_bf16_outputs.json +1026 -0
- train.jsonl +0 -0
- train.py +124 -169
- trainer_log.json +6 -226
- v3-results/adapter/README.md +207 -0
- v3-results/adapter/adapter_config.json +42 -0
- v3-results/adapter/adapter_model.safetensors +3 -0
- v3-results/adapter/additional_chat_templates/tool_use.jinja +152 -0
- v3-results/adapter/chat_template.jinja +5 -0
- v3-results/adapter/special_tokens_map.json +17 -0
- v3-results/adapter/tokenizer.json +3 -0
- v3-results/adapter/tokenizer_config.json +2069 -0
- v3-results/heldout_cases.json +392 -0
- v3-results/train_v3.py +128 -0
- v3-results/training.jsonl +0 -0
- v3-results/v2_quantized_outputs.json +572 -0
- v3-results/v3_quantized_outputs.json +572 -0
- v3-results/v3_reloaded_outputs.json +572 -0
- v3-results/v3_report.json +47 -0
- v3/independent_eval.py +98 -0
- v3/train_v3.py +128 -0
README.md
CHANGED
|
@@ -5,68 +5,40 @@ license: llama3.1
|
|
| 5 |
library_name: transformers
|
| 6 |
pipeline_tag: text-generation
|
| 7 |
base_model: NousResearch/DeepHermes-3-Llama-3-8B-Preview
|
| 8 |
-
base_model_relation: finetune
|
| 9 |
-
model_name: Llama-3.1-Astheral
|
| 10 |
tags:
|
| 11 |
- astheral
|
| 12 |
- llama
|
| 13 |
- instruct
|
| 14 |
-
- reasoning
|
| 15 |
- qlora
|
| 16 |
---
|
|
|
|
| 17 |
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
**Built with Llama.** Astheral is SEE7's 8.03B assistant for instruction following, coding, concise explanations and fictional roleplay.
|
| 21 |
-
|
| 22 |
-
## Instruct and thinking
|
| 23 |
-
|
| 24 |
-
Choose **instruct** for direct answers or **thinking** for a worked solution before the final answer. Both modes use the same trained weights. The included inference script selects the mode and accepts your own system instructions with --system.
|
| 25 |
-
|
| 26 |
-
## What's changed in v2
|
| 27 |
-
|
| 28 |
-
- Balanced training across general instruction tasks, formatting, text editing, Python, and multi-step mathematics.
|
| 29 |
-
- Rank-32 adapters across attention and feed-forward projections.
|
| 30 |
-
- Explicit training to account for all groups in counting problems and preserve requested output structure.
|
| 31 |
-
- Short, complete worked solutions in thinking mode.
|
| 32 |
-
- A saved, merged BF16 checkpoint plus the separately downloadable adapter.
|
| 33 |
-
|
| 34 |
-
Training used **4,080 examples**, **1084 optimizer steps**, and **83,886,080 trainable adapter parameters** on a remote **NVIDIA A10G**. The original DeepHermes weights provide the starting model.
|
| 35 |
|
| 36 |
## Measured results
|
| 37 |
|
| 38 |
-
Both models
|
| 39 |
-
|
| 40 |
-
| Check | DeepHermes base | Astheral v2 |
|
| 41 |
-
|---|---:|---:|
|
| 42 |
-
| Formatting | 9/14 | 14/14 |
|
| 43 |
-
| Numerical reasoning | 16/16 | 16/16 |
|
| 44 |
-
| Executable Python | 4/6 | 6/6 |
|
| 45 |
-
| Core checks total | 29/36 | 36/36 |
|
| 46 |
-
| Known regression cases | 1/2 | 2/2 |
|
| 47 |
-
|
| 48 |
-
The two regression cases cover the alpha/omega line format and counting the remaining red and blue balls together. Full prompts, answers, checker code and run metrics are available in evaluate.py, base_outputs.json, astheral_merged_outputs.json and evaluation.json. Six additional samples cover roleplay, prompt design, authorized networking explanations, uncertainty, debugging and summarization.
|
| 49 |
-
|
| 50 |
-
Default mode formatting was verified in **60/60 responses**: instruct responses omit thinking tags; thinking responses contain a completed thinking block.
|
| 51 |
|
| 52 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
|
| 54 |
-
|
| 55 |
|
| 56 |
-
|
| 57 |
|
| 58 |
-
|
| 59 |
-
python inference.py "A box has 23 red and 17 blue beads; 6 red are removed. How many remain?" --mode thinking
|
| 60 |
-
python inference.py "Write a fictional starship announcement." --mode instruct --load-in-4bit
|
| 61 |
|
| 62 |
-
|
| 63 |
|
| 64 |
-
|
| 65 |
|
| 66 |
-
|
| 67 |
|
| 68 |
-
|
| 69 |
|
| 70 |
-
|
| 71 |
|
| 72 |
-
|
|
|
|
| 5 |
library_name: transformers
|
| 6 |
pipeline_tag: text-generation
|
| 7 |
base_model: NousResearch/DeepHermes-3-Llama-3-8B-Preview
|
|
|
|
|
|
|
| 8 |
tags:
|
| 9 |
- astheral
|
| 10 |
- llama
|
| 11 |
- instruct
|
|
|
|
| 12 |
- qlora
|
| 13 |
---
|
| 14 |
+
# Astheral V3
|
| 15 |
|
| 16 |
+
**Built with Llama.** Real 8.03B merged BF16 weights, fine-tuned from Astheral V2. This is a modest instruction-following update, not a claim of broadly improved reasoning or coding.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
## Measured results
|
| 19 |
|
| 20 |
+
Both final models were evaluated in BF16 with identical prompts and decoding settings.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
+
| Suite | V2 | V3 |
|
| 23 |
+
| --- | --- | --- |
|
| 24 |
+
| Task-family holdout (60 cases) | 41/60 | 43/60 |
|
| 25 |
+
| GSM8K test subset | 6/8 | 6/8 |
|
| 26 |
+
| HumanEval subset | 6/8 | 6/8 |
|
| 27 |
|
| 28 |
+
The 60-case suite uses unseen values within training-related task families; it is not a broad benchmark. Public benchmark subsets are small and not evidence of state-of-the-art performance. Raw prompts, outputs, scoring code and failures are included in paired_bf16_outputs.json, holdout_outputs.json and holdout_evaluation.json.
|
| 29 |
|
| 30 |
+
The quantized adapter scored 49/60 before merging; the shipped BF16 weights scored 43/60 after saving and reloading. The shipped score is the relevant result.
|
| 31 |
|
| 32 |
+
## Training and reproducibility
|
|
|
|
|
|
|
| 33 |
|
| 34 |
+
A10G-small, listed at $1/hour (USD). Rank-16 NF4 QLoRA updates 41,943,040 adapter parameters. Dataset: 1,713 token-filtered examples; 180 optimizer steps, approximately 0.42 epochs. Includes generated arithmetic/formatting/function examples plus retained V2 data. Source and exact data are in v3/ and v3-results/. Reports include the parent revision, seed, timings and memory use.
|
| 35 |
|
| 36 |
+
V2 parent: d9aa899722be4788f57107c2951c72d0ec6614ee. It remains available at tag v2-before-v3.
|
| 37 |
|
| 38 |
+
## Use
|
| 39 |
|
| 40 |
+
Load SEE7/Astheral with Transformers using BF16 on an appropriate GPU. The merged weights require about 16 GB before context/cache overhead. Existing inference.py supports instruct and thinking modes and reads settings.json. For the separate adapter, use the exact V2 parent revision specified in adapter/adapter_config.json, not the V3 merged weights.
|
| 41 |
|
| 42 |
+
## Licensing
|
| 43 |
|
| 44 |
+
Llama 3.1 licensing applies. See LICENSE, UPSTREAM_LLAMA3_LICENSE, NOTICE and the existing data attribution files. Retained V2 examples keep their original source terms; synthetic examples are provided with the reproducible generation script.
|
adapter/adapter_config.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"alpha_pattern": {},
|
| 3 |
"auto_mapping": null,
|
| 4 |
-
"base_model_name_or_path": "
|
| 5 |
"bias": "none",
|
| 6 |
"corda_config": null,
|
| 7 |
"eva_config": null,
|
|
@@ -13,7 +13,7 @@
|
|
| 13 |
"layers_pattern": null,
|
| 14 |
"layers_to_transform": null,
|
| 15 |
"loftq_config": {},
|
| 16 |
-
"lora_alpha":
|
| 17 |
"lora_bias": false,
|
| 18 |
"lora_dropout": 0.05,
|
| 19 |
"megatron_config": null,
|
|
@@ -21,17 +21,17 @@
|
|
| 21 |
"modules_to_save": null,
|
| 22 |
"peft_type": "LORA",
|
| 23 |
"qalora_group_size": 16,
|
| 24 |
-
"r":
|
| 25 |
"rank_pattern": {},
|
| 26 |
-
"revision":
|
| 27 |
"target_modules": [
|
| 28 |
-
"
|
| 29 |
-
"v_proj",
|
| 30 |
"o_proj",
|
| 31 |
-
"k_proj",
|
| 32 |
"up_proj",
|
|
|
|
|
|
|
| 33 |
"q_proj",
|
| 34 |
-
"
|
| 35 |
],
|
| 36 |
"target_parameters": null,
|
| 37 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 1 |
{
|
| 2 |
"alpha_pattern": {},
|
| 3 |
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "SEE7/Astheral",
|
| 5 |
"bias": "none",
|
| 6 |
"corda_config": null,
|
| 7 |
"eva_config": null,
|
|
|
|
| 13 |
"layers_pattern": null,
|
| 14 |
"layers_to_transform": null,
|
| 15 |
"loftq_config": {},
|
| 16 |
+
"lora_alpha": 32,
|
| 17 |
"lora_bias": false,
|
| 18 |
"lora_dropout": 0.05,
|
| 19 |
"megatron_config": null,
|
|
|
|
| 21 |
"modules_to_save": null,
|
| 22 |
"peft_type": "LORA",
|
| 23 |
"qalora_group_size": 16,
|
| 24 |
+
"r": 16,
|
| 25 |
"rank_pattern": {},
|
| 26 |
+
"revision": "d9aa899722be4788f57107c2951c72d0ec6614ee",
|
| 27 |
"target_modules": [
|
| 28 |
+
"gate_proj",
|
|
|
|
| 29 |
"o_proj",
|
|
|
|
| 30 |
"up_proj",
|
| 31 |
+
"down_proj",
|
| 32 |
+
"v_proj",
|
| 33 |
"q_proj",
|
| 34 |
+
"k_proj"
|
| 35 |
],
|
| 36 |
"target_parameters": null,
|
| 37 |
"task_type": "CAUSAL_LM",
|
adapter/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d9a26fcd5590725a2cfdfec6246c93ea91e237e3bf34e214b366b2387feb00ae
|
| 3 |
+
size 167832240
|
artifact_manifest.json
CHANGED
|
@@ -8,8 +8,8 @@
|
|
| 8 |
"sha256": "6b1451bc3455ba759012a8b0d4e5b82becb91c16242e6b365ad249a0a5c4cc98"
|
| 9 |
},
|
| 10 |
"train.jsonl": {
|
| 11 |
-
"bytes":
|
| 12 |
-
"sha256": "
|
| 13 |
},
|
| 14 |
"validation.jsonl": {
|
| 15 |
"bytes": 57424,
|
|
@@ -20,12 +20,12 @@
|
|
| 20 |
"sha256": "820864e76ce5ab18d5dd54a7a4a7419f3b9d5324056b758e976fe28c457f07ef"
|
| 21 |
},
|
| 22 |
"evaluation.json": {
|
| 23 |
-
"bytes":
|
| 24 |
-
"sha256": "
|
| 25 |
},
|
| 26 |
"trainer_log.json": {
|
| 27 |
-
"bytes":
|
| 28 |
-
"sha256": "
|
| 29 |
},
|
| 30 |
"settings.json": {
|
| 31 |
"bytes": 583,
|
|
@@ -44,32 +44,32 @@
|
|
| 44 |
"sha256": "a596868291bbb2f12644a59bb87ca14bd96601247cbc557d4d1c47c606eb81de"
|
| 45 |
},
|
| 46 |
"train.py": {
|
| 47 |
-
"bytes":
|
| 48 |
-
"sha256": "
|
| 49 |
},
|
| 50 |
"config.json": {
|
| 51 |
"bytes": 1043,
|
| 52 |
-
"sha256": "
|
| 53 |
},
|
| 54 |
"generation_config.json": {
|
| 55 |
-
"bytes":
|
| 56 |
-
"sha256": "
|
| 57 |
},
|
| 58 |
"model-00001-of-00005.safetensors": {
|
| 59 |
"bytes": 3986791344,
|
| 60 |
-
"sha256": "
|
| 61 |
},
|
| 62 |
"model-00002-of-00005.safetensors": {
|
| 63 |
"bytes": 3926025416,
|
| 64 |
-
"sha256": "
|
| 65 |
},
|
| 66 |
"model-00003-of-00005.safetensors": {
|
| 67 |
"bytes": 3926025440,
|
| 68 |
-
"sha256": "
|
| 69 |
},
|
| 70 |
"model-00004-of-00005.safetensors": {
|
| 71 |
"bytes": 3171040864,
|
| 72 |
-
"sha256": "
|
| 73 |
},
|
| 74 |
"model-00005-of-00005.safetensors": {
|
| 75 |
"bytes": 1050673280,
|
|
@@ -156,20 +156,20 @@
|
|
| 156 |
"sha256": "fda689d55ae71fc8ca8151b131078b9663843ef965b5d69cad8d2618235edef5"
|
| 157 |
},
|
| 158 |
"README.md": {
|
| 159 |
-
"bytes":
|
| 160 |
-
"sha256": "
|
| 161 |
},
|
| 162 |
"adapter/README.md": {
|
| 163 |
"bytes": 5242,
|
| 164 |
"sha256": "4a1291352e98bd0f745b4d2e6f6a6584428f183f38ba3745d2c1521187bde9d5"
|
| 165 |
},
|
| 166 |
"adapter/adapter_model.safetensors": {
|
| 167 |
-
"bytes":
|
| 168 |
-
"sha256": "
|
| 169 |
},
|
| 170 |
"adapter/adapter_config.json": {
|
| 171 |
-
"bytes":
|
| 172 |
-
"sha256": "
|
| 173 |
},
|
| 174 |
"adapter/chat_template.jinja": {
|
| 175 |
"bytes": 484,
|
|
@@ -200,19 +200,91 @@
|
|
| 200 |
"sha256": "1bc9c9f727f140a1a953b06ebe97a5d0981270b4b2506bb9b58f718d94cf5a1f"
|
| 201 |
},
|
| 202 |
"holdout_evaluation.json": {
|
| 203 |
-
"bytes":
|
| 204 |
-
"sha256": "
|
| 205 |
},
|
| 206 |
"holdout_outputs.json": {
|
| 207 |
-
"bytes":
|
| 208 |
-
"sha256": "
|
| 209 |
},
|
| 210 |
"holdout_eval.py": {
|
| 211 |
-
"bytes":
|
| 212 |
-
"sha256": "
|
| 213 |
},
|
| 214 |
"HUMANEVAL_LICENSE.txt": {
|
| 215 |
"bytes": 1083,
|
| 216 |
"sha256": "bcba3de214851cce46ed5af42d6698044616eeace887c3231bc7a20474ab639e"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 217 |
}
|
| 218 |
}
|
|
|
|
| 8 |
"sha256": "6b1451bc3455ba759012a8b0d4e5b82becb91c16242e6b365ad249a0a5c4cc98"
|
| 9 |
},
|
| 10 |
"train.jsonl": {
|
| 11 |
+
"bytes": 1314989,
|
| 12 |
+
"sha256": "06714a6155a661588cfdfd6bca750b8f34a654475afbf63c00e8794f22cc8a1f"
|
| 13 |
},
|
| 14 |
"validation.jsonl": {
|
| 15 |
"bytes": 57424,
|
|
|
|
| 20 |
"sha256": "820864e76ce5ab18d5dd54a7a4a7419f3b9d5324056b758e976fe28c457f07ef"
|
| 21 |
},
|
| 22 |
"evaluation.json": {
|
| 23 |
+
"bytes": 2408,
|
| 24 |
+
"sha256": "045e35176e473b202b69b3541d38dd1b570d205a794fa1c2ce607253e6cb60af"
|
| 25 |
},
|
| 26 |
"trainer_log.json": {
|
| 27 |
+
"bytes": 226,
|
| 28 |
+
"sha256": "3d60088b5403c9879f2f74a9675ca526133fcbff2fe6195bbfb615607751fa7f"
|
| 29 |
},
|
| 30 |
"settings.json": {
|
| 31 |
"bytes": 583,
|
|
|
|
| 44 |
"sha256": "a596868291bbb2f12644a59bb87ca14bd96601247cbc557d4d1c47c606eb81de"
|
| 45 |
},
|
| 46 |
"train.py": {
|
| 47 |
+
"bytes": 10241,
|
| 48 |
+
"sha256": "29f8c6f7d584eebf637bf99de3e0d50e0d84c1a0d10b2a7c5f8ee69f8b621a40"
|
| 49 |
},
|
| 50 |
"config.json": {
|
| 51 |
"bytes": 1043,
|
| 52 |
+
"sha256": "7a4af87b359d8a181e2a18eaed3c629ca8a95e056d6161c666e9708bf6e54320"
|
| 53 |
},
|
| 54 |
"generation_config.json": {
|
| 55 |
+
"bytes": 231,
|
| 56 |
+
"sha256": "6cba67e3888dfac295499d5660e0f55b8d109d3326659c45fe90314d5cbccbed"
|
| 57 |
},
|
| 58 |
"model-00001-of-00005.safetensors": {
|
| 59 |
"bytes": 3986791344,
|
| 60 |
+
"sha256": "6e4a9bfd682280e0271635d2b4329938f740373ce7d1fe3821d25b210bfbe45f"
|
| 61 |
},
|
| 62 |
"model-00002-of-00005.safetensors": {
|
| 63 |
"bytes": 3926025416,
|
| 64 |
+
"sha256": "ef0783b675748defc6ffc0a64580c578166ac499c5b83b64ceff8d5f60d12801"
|
| 65 |
},
|
| 66 |
"model-00003-of-00005.safetensors": {
|
| 67 |
"bytes": 3926025440,
|
| 68 |
+
"sha256": "30874d39316469ed6bff9cad62a9a576497dc47bc89886fbc67e8454f2733b09"
|
| 69 |
},
|
| 70 |
"model-00004-of-00005.safetensors": {
|
| 71 |
"bytes": 3171040864,
|
| 72 |
+
"sha256": "6fa29d21bcb34fb555d70300cdd2fcb1932bbf2c4c5f2d7a985f0532a1601dce"
|
| 73 |
},
|
| 74 |
"model-00005-of-00005.safetensors": {
|
| 75 |
"bytes": 1050673280,
|
|
|
|
| 156 |
"sha256": "fda689d55ae71fc8ca8151b131078b9663843ef965b5d69cad8d2618235edef5"
|
| 157 |
},
|
| 158 |
"README.md": {
|
| 159 |
+
"bytes": 2287,
|
| 160 |
+
"sha256": "d4c78ae9782ab76e4d2c844c2c23892fd04224f6128c041fbe94bcdd75dbcb15"
|
| 161 |
},
|
| 162 |
"adapter/README.md": {
|
| 163 |
"bytes": 5242,
|
| 164 |
"sha256": "4a1291352e98bd0f745b4d2e6f6a6584428f183f38ba3745d2c1521187bde9d5"
|
| 165 |
},
|
| 166 |
"adapter/adapter_model.safetensors": {
|
| 167 |
+
"bytes": 167832240,
|
| 168 |
+
"sha256": "d9a26fcd5590725a2cfdfec6246c93ea91e237e3bf34e214b366b2387feb00ae"
|
| 169 |
},
|
| 170 |
"adapter/adapter_config.json": {
|
| 171 |
+
"bytes": 963,
|
| 172 |
+
"sha256": "294dc0be03073320633ddcfe2479977bfa92ec382653014e2894fc34eb6045c9"
|
| 173 |
},
|
| 174 |
"adapter/chat_template.jinja": {
|
| 175 |
"bytes": 484,
|
|
|
|
| 200 |
"sha256": "1bc9c9f727f140a1a953b06ebe97a5d0981270b4b2506bb9b58f718d94cf5a1f"
|
| 201 |
},
|
| 202 |
"holdout_evaluation.json": {
|
| 203 |
+
"bytes": 1084,
|
| 204 |
+
"sha256": "fbce8af059fe680606182404058668a327a9cd23dc813423eb60a4ba551fecce"
|
| 205 |
},
|
| 206 |
"holdout_outputs.json": {
|
| 207 |
+
"bytes": 34061,
|
| 208 |
+
"sha256": "9239de420e23dc0646176a7a036c2ba5f2a8987d912a89daba0a492df9e5cb35"
|
| 209 |
},
|
| 210 |
"holdout_eval.py": {
|
| 211 |
+
"bytes": 8577,
|
| 212 |
+
"sha256": "3095c6fcdd834f17990fd9693e4bb45505a0b1532fda6c8f43e440f1fab7d2b5"
|
| 213 |
},
|
| 214 |
"HUMANEVAL_LICENSE.txt": {
|
| 215 |
"bytes": 1083,
|
| 216 |
"sha256": "bcba3de214851cce46ed5af42d6698044616eeace887c3231bc7a20474ab639e"
|
| 217 |
+
},
|
| 218 |
+
"v3-results/adapter/README.md": {
|
| 219 |
+
"bytes": 5180,
|
| 220 |
+
"sha256": "1e72c8245ee442ddfed60cf46879336fd3577234908a6b66ba371809f2e1b025"
|
| 221 |
+
},
|
| 222 |
+
"v3-results/adapter/adapter_config.json": {
|
| 223 |
+
"bytes": 963,
|
| 224 |
+
"sha256": "294dc0be03073320633ddcfe2479977bfa92ec382653014e2894fc34eb6045c9"
|
| 225 |
+
},
|
| 226 |
+
"v3-results/adapter/additional_chat_templates/tool_use.jinja": {
|
| 227 |
+
"bytes": 5004,
|
| 228 |
+
"sha256": "7ce09d55d3690e06c70b5c07228cc7e8c99c43a23cdc027762be4011efcdea6c"
|
| 229 |
+
},
|
| 230 |
+
"v3-results/adapter/chat_template.jinja": {
|
| 231 |
+
"bytes": 484,
|
| 232 |
+
"sha256": "47d5a7b4840c2ff2dfa0828b2465a654bef3a6cdfe032e5e105f395e4ceee812"
|
| 233 |
+
},
|
| 234 |
+
"v3-results/adapter/special_tokens_map.json": {
|
| 235 |
+
"bytes": 340,
|
| 236 |
+
"sha256": "f526dabe70ae46e1fb9caff406987c5e46a18f2c572906efa75b890e7833fb2a"
|
| 237 |
+
},
|
| 238 |
+
"v3-results/adapter/tokenizer.json": {
|
| 239 |
+
"bytes": 17209920,
|
| 240 |
+
"sha256": "6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b"
|
| 241 |
+
},
|
| 242 |
+
"v3-results/adapter/tokenizer_config.json": {
|
| 243 |
+
"bytes": 50720,
|
| 244 |
+
"sha256": "d93d412174451c19c3eaf39974355289d29b34d40114bd6732a688c5e96c197e"
|
| 245 |
+
},
|
| 246 |
+
"v3-results/heldout_cases.json": {
|
| 247 |
+
"bytes": 10608,
|
| 248 |
+
"sha256": "d0d92afbd8b703776effc747f637e942df40e3e906563e75dafddc9530e7576f"
|
| 249 |
+
},
|
| 250 |
+
"v3-results/train_v3.py": {
|
| 251 |
+
"bytes": 10241,
|
| 252 |
+
"sha256": "29f8c6f7d584eebf637bf99de3e0d50e0d84c1a0d10b2a7c5f8ee69f8b621a40"
|
| 253 |
+
},
|
| 254 |
+
"v3-results/training.jsonl": {
|
| 255 |
+
"bytes": 1314989,
|
| 256 |
+
"sha256": "06714a6155a661588cfdfd6bca750b8f34a654475afbf63c00e8794f22cc8a1f"
|
| 257 |
+
},
|
| 258 |
+
"v3-results/v2_quantized_outputs.json": {
|
| 259 |
+
"bytes": 16335,
|
| 260 |
+
"sha256": "465ec17055801e5ad3f6397751fb26aae868fb6d5e5cc68c1b783758d9ad5f18"
|
| 261 |
+
},
|
| 262 |
+
"v3-results/v3_quantized_outputs.json": {
|
| 263 |
+
"bytes": 15253,
|
| 264 |
+
"sha256": "c3ef696a0e0fa15a458a92c2618875d6a7d968588b6cc317c81b13f9217dbd77"
|
| 265 |
+
},
|
| 266 |
+
"v3-results/v3_reloaded_outputs.json": {
|
| 267 |
+
"bytes": 16121,
|
| 268 |
+
"sha256": "2d621e1690e6ec6ddc3d45427e750f32d57e6f60880b95074e36e7028f3ea2d7"
|
| 269 |
+
},
|
| 270 |
+
"v3-results/v3_report.json": {
|
| 271 |
+
"bytes": 1062,
|
| 272 |
+
"sha256": "6c13f805e1f77dc325b0ae5442c548aa9de3abd41ca441f59f39762a44d0587b"
|
| 273 |
+
},
|
| 274 |
+
"v3/independent_eval.py": {
|
| 275 |
+
"bytes": 8577,
|
| 276 |
+
"sha256": "3095c6fcdd834f17990fd9693e4bb45505a0b1532fda6c8f43e440f1fab7d2b5"
|
| 277 |
+
},
|
| 278 |
+
"v3/train_v3.py": {
|
| 279 |
+
"bytes": 10241,
|
| 280 |
+
"sha256": "29f8c6f7d584eebf637bf99de3e0d50e0d84c1a0d10b2a7c5f8ee69f8b621a40"
|
| 281 |
+
},
|
| 282 |
+
"paired_bf16_outputs.json": {
|
| 283 |
+
"bytes": 32202,
|
| 284 |
+
"sha256": "defe6b169178bed42ec5a7127d76c51d06680aa10ddaf60c5dfcfd8a53ffc490"
|
| 285 |
+
},
|
| 286 |
+
"v3-results/adapter/adapter_model.safetensors": {
|
| 287 |
+
"bytes": 167832240,
|
| 288 |
+
"sha256": "d9a26fcd5590725a2cfdfec6246c93ea91e237e3bf34e214b366b2387feb00ae"
|
| 289 |
}
|
| 290 |
}
|
config.json
CHANGED
|
@@ -7,7 +7,7 @@
|
|
| 7 |
"instruct",
|
| 8 |
"thinking"
|
| 9 |
],
|
| 10 |
-
"astheral_version": "
|
| 11 |
"attention_bias": false,
|
| 12 |
"attention_dropout": 0.0,
|
| 13 |
"bos_token_id": 128000,
|
|
|
|
| 7 |
"instruct",
|
| 8 |
"thinking"
|
| 9 |
],
|
| 10 |
+
"astheral_version": "3",
|
| 11 |
"attention_bias": false,
|
| 12 |
"attention_dropout": 0.0,
|
| 13 |
"bos_token_id": 128000,
|
evaluation.json
CHANGED
|
@@ -1,149 +1,97 @@
|
|
| 1 |
{
|
| 2 |
-
"version": "
|
| 3 |
-
"repair_pass": 2,
|
| 4 |
-
"base_model": "NousResearch/DeepHermes-3-Llama-3-8B-Preview",
|
| 5 |
-
"base_revision": "53d902c7dcfa4c749a3b455149064f82af2e2549",
|
| 6 |
-
"initial_adapter_revision": "930a185fd9f4e7fffe44283fa9009ccbd2aa0af7",
|
| 7 |
-
"training_examples": 4080,
|
| 8 |
-
"repair_training_examples": 1617,
|
| 9 |
-
"validation_examples": 64,
|
| 10 |
-
"trainable_parameters": 83886080,
|
| 11 |
-
"rank": 32,
|
| 12 |
-
"target_modules": "all-linear",
|
| 13 |
-
"optimizer_steps": 1084,
|
| 14 |
-
"training_stages": [
|
| 15 |
-
{
|
| 16 |
-
"method": "NF4 QLoRA",
|
| 17 |
-
"examples": 2429,
|
| 18 |
-
"steps": 456,
|
| 19 |
-
"epochs": 1.500617538081515
|
| 20 |
-
},
|
| 21 |
-
{
|
| 22 |
-
"method": "BF16 LoRA",
|
| 23 |
-
"examples": 1714,
|
| 24 |
-
"steps": 323,
|
| 25 |
-
"epochs": 1.5040840140023337
|
| 26 |
-
},
|
| 27 |
-
{
|
| 28 |
-
"method": "BF16 LoRA",
|
| 29 |
-
"examples": 1617,
|
| 30 |
-
"steps": 305,
|
| 31 |
-
"epochs": 1.5046382189239331
|
| 32 |
-
}
|
| 33 |
-
],
|
| 34 |
-
"validation_loss_before": 1.3753070831298828,
|
| 35 |
-
"validation_loss_after": 0.0020501785911619663,
|
| 36 |
-
"base_scores": {
|
| 37 |
-
"regression": {
|
| 38 |
-
"passed": 1,
|
| 39 |
-
"total": 2
|
| 40 |
-
},
|
| 41 |
-
"fresh": {
|
| 42 |
-
"passed": 29,
|
| 43 |
-
"total": 36
|
| 44 |
-
},
|
| 45 |
-
"exact": {
|
| 46 |
-
"passed": 5,
|
| 47 |
-
"total": 10
|
| 48 |
-
},
|
| 49 |
-
"json": {
|
| 50 |
-
"passed": 4,
|
| 51 |
-
"total": 4
|
| 52 |
-
},
|
| 53 |
-
"number": {
|
| 54 |
-
"passed": 16,
|
| 55 |
-
"total": 16
|
| 56 |
-
},
|
| 57 |
-
"code": {
|
| 58 |
-
"passed": 4,
|
| 59 |
-
"total": 6
|
| 60 |
-
}
|
| 61 |
-
},
|
| 62 |
-
"astheral_scores": {
|
| 63 |
-
"regression": {
|
| 64 |
-
"passed": 2,
|
| 65 |
-
"total": 2
|
| 66 |
-
},
|
| 67 |
-
"fresh": {
|
| 68 |
-
"passed": 36,
|
| 69 |
-
"total": 36
|
| 70 |
-
},
|
| 71 |
-
"exact": {
|
| 72 |
-
"passed": 10,
|
| 73 |
-
"total": 10
|
| 74 |
-
},
|
| 75 |
-
"json": {
|
| 76 |
-
"passed": 4,
|
| 77 |
-
"total": 4
|
| 78 |
-
},
|
| 79 |
-
"number": {
|
| 80 |
-
"passed": 16,
|
| 81 |
-
"total": 16
|
| 82 |
-
},
|
| 83 |
-
"code": {
|
| 84 |
-
"passed": 6,
|
| 85 |
-
"total": 6
|
| 86 |
-
},
|
| 87 |
-
"confirmation": {
|
| 88 |
-
"passed": 16,
|
| 89 |
-
"total": 16
|
| 90 |
-
}
|
| 91 |
-
},
|
| 92 |
-
"quality_gate_passed": true,
|
| 93 |
-
"hardware": "NVIDIA A10G",
|
| 94 |
-
"max_gpu_memory_allocated_bytes": 17538217984,
|
| 95 |
"training": {
|
| 96 |
-
"
|
| 97 |
-
"
|
| 98 |
-
"
|
| 99 |
-
"
|
| 100 |
-
"
|
| 101 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
},
|
| 103 |
-
"
|
| 104 |
-
"
|
| 105 |
-
"
|
| 106 |
-
"
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
"
|
| 111 |
-
|
| 112 |
-
"
|
| 113 |
-
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
"
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 137 |
}
|
| 138 |
-
},
|
| 139 |
-
"elapsed_seconds": 1417.947889062998,
|
| 140 |
-
"evaluation_parser_revision": "2: boxed answer, unambiguous leading answer paragraph, then numeric fallback",
|
| 141 |
-
"mode_adherence": {
|
| 142 |
-
"passed": 60,
|
| 143 |
-
"total": 60
|
| 144 |
-
},
|
| 145 |
-
"thinking_output_characters": {
|
| 146 |
-
"base": 1426.7058823529412,
|
| 147 |
-
"astheral": 136.22727272727272
|
| 148 |
}
|
| 149 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"version": "3",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
"training": {
|
| 4 |
+
"status": "complete",
|
| 5 |
+
"parent_revision": "d9aa899722be4788f57107c2951c72d0ec6614ee",
|
| 6 |
+
"hardware": "NVIDIA A10G",
|
| 7 |
+
"job_timeout_seconds": 5400,
|
| 8 |
+
"listed_usd_per_hour": 1.0,
|
| 9 |
+
"training_seed": 20261113,
|
| 10 |
+
"baseline_passed": 40,
|
| 11 |
+
"elapsed_seconds": 622.09,
|
| 12 |
+
"training_examples": 1713,
|
| 13 |
+
"trainable_parameters": 41943040,
|
| 14 |
+
"training": {
|
| 15 |
+
"train_runtime": 390.2404,
|
| 16 |
+
"train_samples_per_second": 1.845,
|
| 17 |
+
"train_steps_per_second": 0.461,
|
| 18 |
+
"total_flos": 3960249019342848.0,
|
| 19 |
+
"train_loss": 0.2867420093880759,
|
| 20 |
+
"epoch": 0.4203152364273205
|
| 21 |
+
},
|
| 22 |
+
"optimizer_steps": 180,
|
| 23 |
+
"candidate_passed": 49,
|
| 24 |
+
"reloaded_passed": 43,
|
| 25 |
+
"groups": {
|
| 26 |
+
"integer": {
|
| 27 |
+
"baseline": 0,
|
| 28 |
+
"candidate": 3,
|
| 29 |
+
"total": 20
|
| 30 |
+
},
|
| 31 |
+
"exact": {
|
| 32 |
+
"baseline": 20,
|
| 33 |
+
"candidate": 20,
|
| 34 |
+
"total": 20
|
| 35 |
+
},
|
| 36 |
+
"code": {
|
| 37 |
+
"baseline": 10,
|
| 38 |
+
"candidate": 10,
|
| 39 |
+
"total": 10
|
| 40 |
+
},
|
| 41 |
+
"json": {
|
| 42 |
+
"baseline": 10,
|
| 43 |
+
"candidate": 10,
|
| 44 |
+
"total": 10
|
| 45 |
+
}
|
| 46 |
+
},
|
| 47 |
+
"quality_gate_passed": true,
|
| 48 |
+
"peak_gpu_bytes": 18991949312
|
| 49 |
},
|
| 50 |
+
"matched_bf16_and_independent_tests": {
|
| 51 |
+
"comparison": "V2 versus V3, both BF16",
|
| 52 |
+
"v2_revision": "d9aa899722be4788f57107c2951c72d0ec6614ee",
|
| 53 |
+
"paired_bf16_scores": {
|
| 54 |
+
"base": 41,
|
| 55 |
+
"astheral": 43
|
| 56 |
+
},
|
| 57 |
+
"suite": "GSM8K test rows 231-238; HumanEval selected tasks 0,3,7,12,18,25,36,45",
|
| 58 |
+
"scores": {
|
| 59 |
+
"base": {
|
| 60 |
+
"code": {
|
| 61 |
+
"passed": 6,
|
| 62 |
+
"total": 8,
|
| 63 |
+
"token_limited": 0
|
| 64 |
+
},
|
| 65 |
+
"number": {
|
| 66 |
+
"passed": 6,
|
| 67 |
+
"total": 8,
|
| 68 |
+
"token_limited": 0
|
| 69 |
+
}
|
| 70 |
+
},
|
| 71 |
+
"astheral": {
|
| 72 |
+
"code": {
|
| 73 |
+
"passed": 6,
|
| 74 |
+
"total": 8,
|
| 75 |
+
"token_limited": 0
|
| 76 |
+
},
|
| 77 |
+
"number": {
|
| 78 |
+
"passed": 6,
|
| 79 |
+
"total": 8,
|
| 80 |
+
"token_limited": 0
|
| 81 |
+
}
|
| 82 |
+
}
|
| 83 |
+
},
|
| 84 |
+
"no_material_regression_gate": true,
|
| 85 |
+
"astheral_revision": "8d553fe747fd6e12fd650c97b712e7009a8b4f9e",
|
| 86 |
+
"settings": {
|
| 87 |
+
"sampling": false,
|
| 88 |
+
"code_max_new_tokens": 768,
|
| 89 |
+
"math_max_new_tokens": 1536
|
| 90 |
+
},
|
| 91 |
+
"hardware": "NVIDIA A10G",
|
| 92 |
+
"source_revisions": {
|
| 93 |
+
"gsm8k": "740312add88f781978c0658806c59bc2815b9866",
|
| 94 |
+
"humaneval": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544"
|
| 95 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
}
|
| 97 |
}
|
generation_config.json
CHANGED
|
@@ -1,17 +1,13 @@
|
|
| 1 |
{
|
| 2 |
"_from_model_config": true,
|
| 3 |
"bos_token_id": 128000,
|
| 4 |
-
"do_sample": false,
|
| 5 |
"eos_token_id": [
|
| 6 |
128001,
|
| 7 |
128008,
|
| 8 |
128009
|
| 9 |
],
|
| 10 |
"max_length": 131072,
|
|
|
|
| 11 |
"pad_token_id": 128004,
|
| 12 |
-
"
|
| 13 |
-
|
| 14 |
-
"transformers_version": "4.56.2",
|
| 15 |
-
"top_k": 50,
|
| 16 |
-
"max_new_tokens": 1024
|
| 17 |
-
}
|
|
|
|
| 1 |
{
|
| 2 |
"_from_model_config": true,
|
| 3 |
"bos_token_id": 128000,
|
|
|
|
| 4 |
"eos_token_id": [
|
| 5 |
128001,
|
| 6 |
128008,
|
| 7 |
128009
|
| 8 |
],
|
| 9 |
"max_length": 131072,
|
| 10 |
+
"max_new_tokens": 1024,
|
| 11 |
"pad_token_id": 128004,
|
| 12 |
+
"transformers_version": "4.56.2"
|
| 13 |
+
}
|
|
|
|
|
|
|
|
|
|
|
|
holdout_eval.py
CHANGED
|
@@ -1,13 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
| 1 |
"""Independent GSM8K and HumanEval subset comparison on remote GPUs."""
|
| 2 |
import json,re,sys,subprocess,tempfile,ast,urllib.request,urllib.parse,gc,hashlib
|
| 3 |
from pathlib import Path
|
| 4 |
import torch
|
| 5 |
from transformers import AutoModelForCausalLM,AutoTokenizer
|
| 6 |
from huggingface_hub import HfApi,hf_hub_download,CommitOperationAdd
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
| 8 |
api=HfApi();REPO="SEE7/Astheral";BASE="NousResearch/DeepHermes-3-Llama-3-8B-Preview";REV="53d902c7dcfa4c749a3b455149064f82af2e2549"
|
| 9 |
ROOT=Path(__file__).resolve().parent
|
| 10 |
-
|
|
|
|
| 11 |
settings=json.loads(Path(hf_hub_download(REPO,"settings.json",revision=revision)).read_text())
|
| 12 |
def rows(repo,config,split,offset,length):
|
| 13 |
q=urllib.parse.urlencode(dict(dataset=repo,config=config,split=split,offset=offset,length=length))
|
|
@@ -37,13 +44,26 @@ def check_code(c,text):
|
|
| 37 |
with tempfile.TemporaryDirectory() as d:
|
| 38 |
p=Path(d)/"check.py"
|
| 39 |
p.write_text("import resource\nresource.setrlimit(resource.RLIMIT_CPU,(3,3))\nresource.setrlimit(resource.RLIMIT_AS,(536870912,536870912))\n"+code+"\n"+c["tests"])
|
| 40 |
-
r=subprocess.run([sys.executable,"-I",str(p)],cwd=d,timeout=5,capture_output=True)
|
| 41 |
return r.returncode==0
|
| 42 |
except Exception:return False
|
| 43 |
-
all_results={};scores={}
|
| 44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
tok=AutoTokenizer.from_pretrained(repo,revision=rev);tok.pad_token_id=128004;tok.padding_side="left"
|
| 46 |
model=AutoModelForCausalLM.from_pretrained(repo,revision=rev,torch_dtype=torch.bfloat16,device_map={"":"cuda:0"},attn_implementation="sdpa").eval()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
result=[]
|
| 48 |
with torch.inference_mode():
|
| 49 |
for mode in ["instruct","thinking"]:
|
|
@@ -66,12 +86,13 @@ for tag,repo,rev in [("base",BASE,REV),("astheral",REPO,revision)]:
|
|
| 66 |
chosen_rows=[r for r in result if r["kind"]==kind]
|
| 67 |
scores[tag][kind]={"passed":sum(r["pass"] for r in chosen_rows),"total":len(chosen_rows),"token_limited":sum(r["hit_token_limit"] for r in chosen_rows)}
|
| 68 |
del model;gc.collect();torch.cuda.empty_cache()
|
| 69 |
-
passed=sum(v["passed"] for v in scores["astheral"].values())>=sum(v["passed"] for v in scores["base"].values()) and all(scores["astheral"][k]["passed"]>=scores["base"][k]["passed"]
|
| 70 |
-
|
| 71 |
-
|
|
|
|
| 72 |
manifest=json.loads(Path(hf_hub_download(REPO,"artifact_manifest.json")).read_text())
|
| 73 |
for n,b in changed.items():manifest[n]={"bytes":len(b.encode()),"sha256":hashlib.sha256(b.encode()).hexdigest()}
|
| 74 |
changed["artifact_manifest.json"]=json.dumps(manifest,indent=2)
|
| 75 |
-
api.create_commit(repo_id=REPO,operations=[CommitOperationAdd(path_in_repo=n,path_or_fileobj=b.encode()) for n,b in changed.items()],commit_message="Record
|
| 76 |
print("HOLDOUT_REPORT",json.dumps(report),flush=True)
|
| 77 |
if not passed:raise SystemExit(5)
|
|
|
|
| 1 |
+
# /// script
|
| 2 |
+
# dependencies = ["torch==2.8.0", "transformers==4.56.2", "huggingface_hub==0.34.4", "accelerate==1.10.1", "safetensors==0.6.2"]
|
| 3 |
+
# ///
|
| 4 |
"""Independent GSM8K and HumanEval subset comparison on remote GPUs."""
|
| 5 |
import json,re,sys,subprocess,tempfile,ast,urllib.request,urllib.parse,gc,hashlib
|
| 6 |
from pathlib import Path
|
| 7 |
import torch
|
| 8 |
from transformers import AutoModelForCausalLM,AutoTokenizer
|
| 9 |
from huggingface_hub import HfApi,hf_hub_download,CommitOperationAdd
|
| 10 |
+
def task_check(case,text):
|
| 11 |
+
final=text.split("</think>")[-1]
|
| 12 |
+
numbers=re.findall(r"-?\d+(?:\.\d+)?",final.replace(",",""))
|
| 13 |
+
return bool(numbers) and abs(float(numbers[-1])-float(case["expected"]))<1e-6
|
| 14 |
api=HfApi();REPO="SEE7/Astheral";BASE="NousResearch/DeepHermes-3-Llama-3-8B-Preview";REV="53d902c7dcfa4c749a3b455149064f82af2e2549"
|
| 15 |
ROOT=Path(__file__).resolve().parent
|
| 16 |
+
BRANCH="v3-candidate-20261113"
|
| 17 |
+
revision=api.model_info(REPO,revision=BRANCH).sha
|
| 18 |
settings=json.loads(Path(hf_hub_download(REPO,"settings.json",revision=revision)).read_text())
|
| 19 |
def rows(repo,config,split,offset,length):
|
| 20 |
q=urllib.parse.urlencode(dict(dataset=repo,config=config,split=split,offset=offset,length=length))
|
|
|
|
| 44 |
with tempfile.TemporaryDirectory() as d:
|
| 45 |
p=Path(d)/"check.py"
|
| 46 |
p.write_text("import resource\nresource.setrlimit(resource.RLIMIT_CPU,(3,3))\nresource.setrlimit(resource.RLIMIT_AS,(536870912,536870912))\n"+code+"\n"+c["tests"])
|
| 47 |
+
r=subprocess.run([sys.executable,"-I",str(p)],cwd=d,timeout=5,capture_output=True,env={"PATH":"/usr/bin:/bin"})
|
| 48 |
return r.returncode==0
|
| 49 |
except Exception:return False
|
| 50 |
+
all_results={};scores={};paired_results={}
|
| 51 |
+
source=Path(hf_hub_download(REPO,'v3/train_v3.py',revision=revision)).read_text()
|
| 52 |
+
grade_node=next(n for n in ast.parse(source).body if isinstance(n,ast.FunctionDef) and n.name=='grade')
|
| 53 |
+
grade_scope={'json':json,'re':re};exec(compile(ast.Module(body=[grade_node],type_ignores=[]),'grader','exec'),grade_scope)
|
| 54 |
+
paired_cases=json.loads(Path(hf_hub_download(REPO,'v3-results/heldout_cases.json',revision=revision)).read_text())
|
| 55 |
+
for tag,repo,rev in [("base",REPO,"d9aa899722be4788f57107c2951c72d0ec6614ee"),("astheral",REPO,revision)]:
|
| 56 |
tok=AutoTokenizer.from_pretrained(repo,revision=rev);tok.pad_token_id=128004;tok.padding_side="left"
|
| 57 |
model=AutoModelForCausalLM.from_pretrained(repo,revision=rev,torch_dtype=torch.bfloat16,device_map={"":"cuda:0"},attn_implementation="sdpa").eval()
|
| 58 |
+
paired=[]
|
| 59 |
+
with torch.inference_mode():
|
| 60 |
+
for c in paired_cases:
|
| 61 |
+
inp=tok.apply_chat_template([{"role":"system","content":settings["system"]},{"role":"user","content":c["prompt"]}],tokenize=True,add_generation_prompt=True,return_dict=True,return_tensors="pt").to("cuda")
|
| 62 |
+
output=model.generate(**inp,max_new_tokens=96,do_sample=False,pad_token_id=128004,eos_token_id=128009)
|
| 63 |
+
text=tok.decode(output[0,inp["input_ids"].shape[1]:],skip_special_tokens=True)
|
| 64 |
+
paired.append({**c,"output":text,"pass":grade_scope["grade"](c,text)})
|
| 65 |
+
print("BF16_PAIRED",tag,c["id"],paired[-1]["pass"],flush=True)
|
| 66 |
+
paired_results[tag]=paired
|
| 67 |
result=[]
|
| 68 |
with torch.inference_mode():
|
| 69 |
for mode in ["instruct","thinking"]:
|
|
|
|
| 86 |
chosen_rows=[r for r in result if r["kind"]==kind]
|
| 87 |
scores[tag][kind]={"passed":sum(r["pass"] for r in chosen_rows),"total":len(chosen_rows),"token_limited":sum(r["hit_token_limit"] for r in chosen_rows)}
|
| 88 |
del model;gc.collect();torch.cuda.empty_cache()
|
| 89 |
+
passed=sum(v["passed"] for v in scores["astheral"].values())>=sum(v["passed"] for v in scores["base"].values()) and all(scores["astheral"][k]["passed"]>=scores["base"][k]["passed"] for k in ["code","number"])
|
| 90 |
+
passed=passed and sum(r["pass"] for r in paired_results["astheral"])>=sum(r["pass"] for r in paired_results["base"])
|
| 91 |
+
report={"comparison":"V2 versus V3, both BF16","v2_revision":"d9aa899722be4788f57107c2951c72d0ec6614ee","paired_bf16_scores":{k:sum(r["pass"] for r in v) for k,v in paired_results.items()},"suite":"GSM8K test rows 231-238; HumanEval selected tasks 0,3,7,12,18,25,36,45","scores":scores,"no_material_regression_gate":passed,"astheral_revision":revision,"settings":{"sampling":False,"code_max_new_tokens":768,"math_max_new_tokens":1536},"hardware":torch.cuda.get_device_name(0),"source_revisions":{"gsm8k":math_revision,"humaneval":code_revision}}
|
| 92 |
+
changed={"paired_bf16_outputs.json":json.dumps(paired_results,indent=2),"holdout_evaluation.json":json.dumps(report,indent=2),"holdout_outputs.json":json.dumps(all_results,indent=2),"holdout_eval.py":Path(__file__).read_text(),"HUMANEVAL_LICENSE.txt":urllib.request.urlopen("https://raw.githubusercontent.com/openai/human-eval/master/LICENSE",timeout=60).read().decode()}
|
| 93 |
manifest=json.loads(Path(hf_hub_download(REPO,"artifact_manifest.json")).read_text())
|
| 94 |
for n,b in changed.items():manifest[n]={"bytes":len(b.encode()),"sha256":hashlib.sha256(b.encode()).hexdigest()}
|
| 95 |
changed["artifact_manifest.json"]=json.dumps(manifest,indent=2)
|
| 96 |
+
api.create_commit(repo_id=REPO,revision=BRANCH,operations=[CommitOperationAdd(path_in_repo=n,path_or_fileobj=b.encode()) for n,b in changed.items()],commit_message="Record measured V2 versus V3 GSM8K and HumanEval comparison")
|
| 97 |
print("HOLDOUT_REPORT",json.dumps(report),flush=True)
|
| 98 |
if not passed:raise SystemExit(5)
|
holdout_evaluation.json
CHANGED
|
@@ -1,16 +1,22 @@
|
|
| 1 |
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
"suite": "GSM8K test rows 231-238; HumanEval selected tasks 0,3,7,12,18,25,36,45",
|
| 3 |
"scores": {
|
| 4 |
"base": {
|
| 5 |
"code": {
|
| 6 |
-
"passed":
|
| 7 |
"total": 8,
|
| 8 |
"token_limited": 0
|
| 9 |
},
|
| 10 |
"number": {
|
| 11 |
-
"passed":
|
| 12 |
"total": 8,
|
| 13 |
-
"token_limited":
|
| 14 |
}
|
| 15 |
},
|
| 16 |
"astheral": {
|
|
@@ -26,8 +32,8 @@
|
|
| 26 |
}
|
| 27 |
}
|
| 28 |
},
|
| 29 |
-
"no_material_regression_gate":
|
| 30 |
-
"astheral_revision": "
|
| 31 |
"settings": {
|
| 32 |
"sampling": false,
|
| 33 |
"code_max_new_tokens": 768,
|
|
|
|
| 1 |
{
|
| 2 |
+
"comparison": "V2 versus V3, both BF16",
|
| 3 |
+
"v2_revision": "d9aa899722be4788f57107c2951c72d0ec6614ee",
|
| 4 |
+
"paired_bf16_scores": {
|
| 5 |
+
"base": 41,
|
| 6 |
+
"astheral": 43
|
| 7 |
+
},
|
| 8 |
"suite": "GSM8K test rows 231-238; HumanEval selected tasks 0,3,7,12,18,25,36,45",
|
| 9 |
"scores": {
|
| 10 |
"base": {
|
| 11 |
"code": {
|
| 12 |
+
"passed": 6,
|
| 13 |
"total": 8,
|
| 14 |
"token_limited": 0
|
| 15 |
},
|
| 16 |
"number": {
|
| 17 |
+
"passed": 6,
|
| 18 |
"total": 8,
|
| 19 |
+
"token_limited": 0
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"astheral": {
|
|
|
|
| 32 |
}
|
| 33 |
}
|
| 34 |
},
|
| 35 |
+
"no_material_regression_gate": true,
|
| 36 |
+
"astheral_revision": "8d553fe747fd6e12fd650c97b712e7009a8b4f9e",
|
| 37 |
"settings": {
|
| 38 |
"sampling": false,
|
| 39 |
"code_max_new_tokens": 768,
|
holdout_outputs.json
CHANGED
|
@@ -9,7 +9,7 @@
|
|
| 9 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([1.0, 2.0, 3.9, 4.0, 5.0, 2.2], 0.3) == True\n assert candidate([1.0, 2.0, 3.9, 4.0, 5.0, 2.2], 0.05) == False\n assert candidate([1.0, 2.0, 5.9, 4.0, 5.0], 0.95) == True\n assert candidate([1.0, 2.0, 5.9, 4.0, 5.0], 0.8) == False\n assert candidate([1.0, 2.0, 3.0, 4.0, 5.0, 2.0], 0.1) == True\n assert candidate([1.1, 2.2, 3.1, 4.1, 5.1], 1.0) == True\n assert candidate([1.1, 2.2, 3.1, 4.1, 5.1], 0.5) == False\n\n\ncheck(has_close_elements)",
|
| 10 |
"source": "openai/openai_humaneval",
|
| 11 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 12 |
-
"output": "
|
| 13 |
"pass": true,
|
| 14 |
"hit_token_limit": false
|
| 15 |
},
|
|
@@ -22,7 +22,7 @@
|
|
| 22 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([]) == False\n assert candidate([1, 2, -3, 1, 2, -3]) == False\n assert candidate([1, 2, -4, 5, 6]) == True\n assert candidate([1, -1, 2, -2, 5, -5, 4, -4]) == False\n assert candidate([1, -1, 2, -2, 5, -5, 4, -5]) == True\n assert candidate([1, -2, 2, -2, 5, -5, 4, -4]) == True\n\ncheck(below_zero)",
|
| 23 |
"source": "openai/openai_humaneval",
|
| 24 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 25 |
-
"output": "
|
| 26 |
"pass": true,
|
| 27 |
"hit_token_limit": false
|
| 28 |
},
|
|
@@ -35,7 +35,7 @@
|
|
| 35 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([], 'john') == []\n assert candidate(['xxx', 'asd', 'xxy', 'john doe', 'xxxAAA', 'xxx'], 'xxx') == ['xxx', 'xxxAAA', 'xxx']\n assert candidate(['xxx', 'asd', 'aaaxxy', 'john doe', 'xxxAAA', 'xxx'], 'xx') == ['xxx', 'aaaxxy', 'xxxAAA', 'xxx']\n assert candidate(['grunt', 'trumpet', 'prune', 'gruesome'], 'run') == ['grunt', 'prune']\n\ncheck(filter_by_substring)",
|
| 36 |
"source": "openai/openai_humaneval",
|
| 37 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 38 |
-
"output": "
|
| 39 |
"pass": true,
|
| 40 |
"hit_token_limit": false
|
| 41 |
},
|
|
@@ -48,7 +48,7 @@
|
|
| 48 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([]) == None\n assert candidate(['x', 'y', 'z']) == 'x'\n assert candidate(['x', 'yyy', 'zzzz', 'www', 'kkkk', 'abc']) == 'zzzz'\n\ncheck(longest)",
|
| 49 |
"source": "openai/openai_humaneval",
|
| 50 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 51 |
-
"output": "
|
| 52 |
"pass": true,
|
| 53 |
"hit_token_limit": false
|
| 54 |
},
|
|
@@ -61,7 +61,7 @@
|
|
| 61 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate('', 'x') == 0\n assert candidate('xyxyxyx', 'x') == 4\n assert candidate('cacacacac', 'cac') == 4\n assert candidate('john doe', 'john') == 1\n\ncheck(how_many_times)",
|
| 62 |
"source": "openai/openai_humaneval",
|
| 63 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 64 |
-
"output": "
|
| 65 |
"pass": false,
|
| 66 |
"hit_token_limit": false
|
| 67 |
},
|
|
@@ -74,7 +74,7 @@
|
|
| 74 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate(2) == [2]\n assert candidate(4) == [2, 2]\n assert candidate(8) == [2, 2, 2]\n assert candidate(3 * 19) == [3, 19]\n assert candidate(3 * 19 * 3 * 19) == [3, 3, 19, 19]\n assert candidate(3 * 19 * 3 * 19 * 3 * 19) == [3, 3, 3, 19, 19, 19]\n assert candidate(3 * 19 * 19 * 19) == [3, 19, 19, 19]\n assert candidate(3 * 2 * 3) == [2, 3, 3]\n\ncheck(factorize)",
|
| 75 |
"source": "openai/openai_humaneval",
|
| 76 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 77 |
-
"output": "
|
| 78 |
"pass": true,
|
| 79 |
"hit_token_limit": false
|
| 80 |
},
|
|
@@ -87,8 +87,8 @@
|
|
| 87 |
"tests": "\n\nMETADATA = {}\n\n\ndef check(candidate):\n assert candidate(50) == 0\n assert candidate(78) == 2\n assert candidate(79) == 3\n assert candidate(100) == 3\n assert candidate(200) == 6\n assert candidate(4000) == 192\n assert candidate(10000) == 639\n assert candidate(100000) == 8026\n\n\ncheck(fizz_buzz)",
|
| 88 |
"source": "openai/openai_humaneval",
|
| 89 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 90 |
-
"output": "
|
| 91 |
-
"pass":
|
| 92 |
"hit_token_limit": false
|
| 93 |
},
|
| 94 |
{
|
|
@@ -100,7 +100,7 @@
|
|
| 100 |
"tests": "\n\nMETADATA = {}\n\n\ndef check(candidate):\n assert candidate(5, 3) == 7.5\n assert candidate(2, 2) == 2.0\n assert candidate(10, 8) == 40.0\n\n\ncheck(triangle_area)",
|
| 101 |
"source": "openai/openai_humaneval",
|
| 102 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 103 |
-
"output": "
|
| 104 |
"pass": true,
|
| 105 |
"hit_token_limit": false
|
| 106 |
},
|
|
@@ -112,7 +112,7 @@
|
|
| 112 |
"expected": 50.0,
|
| 113 |
"source": "openai/gsm8k",
|
| 114 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 115 |
-
"output": "<think>
|
| 116 |
"pass": true,
|
| 117 |
"hit_token_limit": false
|
| 118 |
},
|
|
@@ -124,7 +124,7 @@
|
|
| 124 |
"expected": 75.0,
|
| 125 |
"source": "openai/gsm8k",
|
| 126 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 127 |
-
"output": "<think>
|
| 128 |
"pass": true,
|
| 129 |
"hit_token_limit": false
|
| 130 |
},
|
|
@@ -136,7 +136,7 @@
|
|
| 136 |
"expected": 12.0,
|
| 137 |
"source": "openai/gsm8k",
|
| 138 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 139 |
-
"output": "<think>
|
| 140 |
"pass": true,
|
| 141 |
"hit_token_limit": false
|
| 142 |
},
|
|
@@ -148,8 +148,8 @@
|
|
| 148 |
"expected": 21.0,
|
| 149 |
"source": "openai/gsm8k",
|
| 150 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 151 |
-
"output": "<think>
|
| 152 |
-
"pass":
|
| 153 |
"hit_token_limit": false
|
| 154 |
},
|
| 155 |
{
|
|
@@ -160,7 +160,7 @@
|
|
| 160 |
"expected": 10.0,
|
| 161 |
"source": "openai/gsm8k",
|
| 162 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 163 |
-
"output": "<think>
|
| 164 |
"pass": true,
|
| 165 |
"hit_token_limit": false
|
| 166 |
},
|
|
@@ -172,9 +172,9 @@
|
|
| 172 |
"expected": 31.0,
|
| 173 |
"source": "openai/gsm8k",
|
| 174 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 175 |
-
"output": "<think>
|
| 176 |
"pass": false,
|
| 177 |
-
"hit_token_limit":
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"id": "gsm8k-test-237",
|
|
@@ -184,7 +184,7 @@
|
|
| 184 |
"expected": 90.0,
|
| 185 |
"source": "openai/gsm8k",
|
| 186 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 187 |
-
"output": "<think>
|
| 188 |
"pass": true,
|
| 189 |
"hit_token_limit": false
|
| 190 |
},
|
|
@@ -196,7 +196,7 @@
|
|
| 196 |
"expected": 68.0,
|
| 197 |
"source": "openai/gsm8k",
|
| 198 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 199 |
-
"output": "<think>
|
| 200 |
"pass": true,
|
| 201 |
"hit_token_limit": false
|
| 202 |
}
|
|
@@ -326,7 +326,7 @@
|
|
| 326 |
"expected": 75.0,
|
| 327 |
"source": "openai/gsm8k",
|
| 328 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 329 |
-
"output": "<think>There were 50/2 = 25 bugs in the garden.\nThe total number of insects in the garden is 25
|
| 330 |
"pass": true,
|
| 331 |
"hit_token_limit": false
|
| 332 |
},
|
|
@@ -338,7 +338,7 @@
|
|
| 338 |
"expected": 12.0,
|
| 339 |
"source": "openai/gsm8k",
|
| 340 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 341 |
-
"output": "<think>There are 36/3 = 12 penguins that jump in the ocean.\nThere are
|
| 342 |
"pass": true,
|
| 343 |
"hit_token_limit": false
|
| 344 |
},
|
|
@@ -374,7 +374,7 @@
|
|
| 374 |
"expected": 31.0,
|
| 375 |
"source": "openai/gsm8k",
|
| 376 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 377 |
-
"output": "<think>First find the total amount of ink he gets from the 25 pens: 25 pens * 5 pens/pens = 125 pens\nThen divide that number by the number of pens he can make from one set of
|
| 378 |
"pass": false,
|
| 379 |
"hit_token_limit": false
|
| 380 |
},
|
|
|
|
| 9 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([1.0, 2.0, 3.9, 4.0, 5.0, 2.2], 0.3) == True\n assert candidate([1.0, 2.0, 3.9, 4.0, 5.0, 2.2], 0.05) == False\n assert candidate([1.0, 2.0, 5.9, 4.0, 5.0], 0.95) == True\n assert candidate([1.0, 2.0, 5.9, 4.0, 5.0], 0.8) == False\n assert candidate([1.0, 2.0, 3.0, 4.0, 5.0, 2.0], 0.1) == True\n assert candidate([1.1, 2.2, 3.1, 4.1, 5.1], 1.0) == True\n assert candidate([1.1, 2.2, 3.1, 4.1, 5.1], 0.5) == False\n\n\ncheck(has_close_elements)",
|
| 10 |
"source": "openai/openai_humaneval",
|
| 11 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 12 |
+
"output": "```python\nfrom typing import List\n\ndef has_close_elements(numbers: List[float], threshold: float) -> bool:\n for i in range(len(numbers)):\n for j in range(i + 1, len(numbers)):\n if abs(numbers[i] - numbers[j]) < threshold:\n return True\n return False\n```",
|
| 13 |
"pass": true,
|
| 14 |
"hit_token_limit": false
|
| 15 |
},
|
|
|
|
| 22 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([]) == False\n assert candidate([1, 2, -3, 1, 2, -3]) == False\n assert candidate([1, 2, -4, 5, 6]) == True\n assert candidate([1, -1, 2, -2, 5, -5, 4, -4]) == False\n assert candidate([1, -1, 2, -2, 5, -5, 4, -5]) == True\n assert candidate([1, -2, 2, -2, 5, -5, 4, -4]) == True\n\ncheck(below_zero)",
|
| 23 |
"source": "openai/openai_humaneval",
|
| 24 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 25 |
+
"output": "```python\nfrom typing import List\n\ndef below_zero(operations: List[int]) -> bool:\n balance = 0\n for operation in operations:\n balance += operation\n if balance < 0:\n return True\n return False\n```",
|
| 26 |
"pass": true,
|
| 27 |
"hit_token_limit": false
|
| 28 |
},
|
|
|
|
| 35 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([], 'john') == []\n assert candidate(['xxx', 'asd', 'xxy', 'john doe', 'xxxAAA', 'xxx'], 'xxx') == ['xxx', 'xxxAAA', 'xxx']\n assert candidate(['xxx', 'asd', 'aaaxxy', 'john doe', 'xxxAAA', 'xxx'], 'xx') == ['xxx', 'aaaxxy', 'xxxAAA', 'xxx']\n assert candidate(['grunt', 'trumpet', 'prune', 'gruesome'], 'run') == ['grunt', 'prune']\n\ncheck(filter_by_substring)",
|
| 36 |
"source": "openai/openai_humaneval",
|
| 37 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 38 |
+
"output": "```python\nfrom typing import List\n\ndef filter_by_substring(strings: List[str], substring: str) -> List[str]:\n return [s for s in strings if substring in s]\n```",
|
| 39 |
"pass": true,
|
| 40 |
"hit_token_limit": false
|
| 41 |
},
|
|
|
|
| 48 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate([]) == None\n assert candidate(['x', 'y', 'z']) == 'x'\n assert candidate(['x', 'yyy', 'zzzz', 'www', 'kkkk', 'abc']) == 'zzzz'\n\ncheck(longest)",
|
| 49 |
"source": "openai/openai_humaneval",
|
| 50 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 51 |
+
"output": "```python\nfrom typing import List, Optional\n\ndef longest(strings: List[str]) -> Optional[str]:\n if not strings:\n return None\n return max(strings, key=len)\n```",
|
| 52 |
"pass": true,
|
| 53 |
"hit_token_limit": false
|
| 54 |
},
|
|
|
|
| 61 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate('', 'x') == 0\n assert candidate('xyxyxyx', 'x') == 4\n assert candidate('cacacacac', 'cac') == 4\n assert candidate('john doe', 'john') == 1\n\ncheck(how_many_times)",
|
| 62 |
"source": "openai/openai_humaneval",
|
| 63 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 64 |
+
"output": "```python\ndef how_many_times(string: str, substring: str) -> int:\n return string.count(substring)\n```",
|
| 65 |
"pass": false,
|
| 66 |
"hit_token_limit": false
|
| 67 |
},
|
|
|
|
| 74 |
"tests": "\n\nMETADATA = {\n 'author': 'jt',\n 'dataset': 'test'\n}\n\n\ndef check(candidate):\n assert candidate(2) == [2]\n assert candidate(4) == [2, 2]\n assert candidate(8) == [2, 2, 2]\n assert candidate(3 * 19) == [3, 19]\n assert candidate(3 * 19 * 3 * 19) == [3, 3, 19, 19]\n assert candidate(3 * 19 * 3 * 19 * 3 * 19) == [3, 3, 3, 19, 19, 19]\n assert candidate(3 * 19 * 19 * 19) == [3, 19, 19, 19]\n assert candidate(3 * 2 * 3) == [2, 3, 3]\n\ncheck(factorize)",
|
| 75 |
"source": "openai/openai_humaneval",
|
| 76 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 77 |
+
"output": "```python\nfrom typing import List\n\ndef factorize(n: int) -> List[int]:\n factors = []\n divisor = 2\n while divisor * divisor <= n:\n if n % divisor == 0:\n factors.append(divisor)\n n //= divisor\n else:\n divisor += 1\n if n > 1:\n factors.append(n)\n return factors\n```",
|
| 78 |
"pass": true,
|
| 79 |
"hit_token_limit": false
|
| 80 |
},
|
|
|
|
| 87 |
"tests": "\n\nMETADATA = {}\n\n\ndef check(candidate):\n assert candidate(50) == 0\n assert candidate(78) == 2\n assert candidate(79) == 3\n assert candidate(100) == 3\n assert candidate(200) == 6\n assert candidate(4000) == 192\n assert candidate(10000) == 639\n assert candidate(100000) == 8026\n\n\ncheck(fizz_buzz)",
|
| 88 |
"source": "openai/openai_humaneval",
|
| 89 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 90 |
+
"output": "```python\nfrom itertools import count\n\ndef fizz_buzz(n: int) -> int:\n return sum(1 for i in count(1, 1) if i < n and (i % 11 == 0 or i % 13 == 0) and str(i).count('7'))\n```",
|
| 91 |
+
"pass": false,
|
| 92 |
"hit_token_limit": false
|
| 93 |
},
|
| 94 |
{
|
|
|
|
| 100 |
"tests": "\n\nMETADATA = {}\n\n\ndef check(candidate):\n assert candidate(5, 3) == 7.5\n assert candidate(2, 2) == 2.0\n assert candidate(10, 8) == 40.0\n\n\ncheck(triangle_area)",
|
| 101 |
"source": "openai/openai_humaneval",
|
| 102 |
"source_revision": "7dce6050a7d6d172f3cc5c32aa97f52fa1a2e544",
|
| 103 |
+
"output": "```python\nimport math\n\ndef triangle_area(a, h):\n return 0.5 * a * h\n```",
|
| 104 |
"pass": true,
|
| 105 |
"hit_token_limit": false
|
| 106 |
},
|
|
|
|
| 112 |
"expected": 50.0,
|
| 113 |
"source": "openai/gsm8k",
|
| 114 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 115 |
+
"output": "<think>On Monday, Kylie collects 5 + 20 = 25 shells.\nOn Tuesday, Kylie collects 2 * 25 = 50 shells.</think>\nAnswer: 50",
|
| 116 |
"pass": true,
|
| 117 |
"hit_token_limit": false
|
| 118 |
},
|
|
|
|
| 124 |
"expected": 75.0,
|
| 125 |
"source": "openai/gsm8k",
|
| 126 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 127 |
+
"output": "<think>There were 50/2 = 25 bugs in the garden.\nThe total number of insects in the garden is 25 + 50 = 75.</think>\nAnswer: 75",
|
| 128 |
"pass": true,
|
| 129 |
"hit_token_limit": false
|
| 130 |
},
|
|
|
|
| 136 |
"expected": 12.0,
|
| 137 |
"source": "openai/gsm8k",
|
| 138 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 139 |
+
"output": "<think>There are 36/3 = 12 penguins that jump in the ocean.\nThere are also 36/3 = 12 penguins that go inside the cave.\nSo, 12 + 12 = 24 penguins are no longer sunbathing.\nTherefore, 36 - 24 = 12 penguins are still sunbathing.</think>\nAnswer: 12",
|
| 140 |
"pass": true,
|
| 141 |
"hit_token_limit": false
|
| 142 |
},
|
|
|
|
| 148 |
"expected": 21.0,
|
| 149 |
"source": "openai/gsm8k",
|
| 150 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 151 |
+
"output": "<think>He eats 6 x 2 = 12 tomatoes per week\nHe needs 12 / 3 = 4 vines</think>\nAnswer: 4",
|
| 152 |
+
"pass": false,
|
| 153 |
"hit_token_limit": false
|
| 154 |
},
|
| 155 |
{
|
|
|
|
| 160 |
"expected": 10.0,
|
| 161 |
"source": "openai/gsm8k",
|
| 162 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 163 |
+
"output": "<think>First find the total number of slices needed by multiplying the number of friends by the number of slices each wants: 20 * 4 = 80\nThen divide that by the number of slices per pizza to find the number of pizzas needed: 80 / 8 = 10</think>\nAnswer: 10",
|
| 164 |
"pass": true,
|
| 165 |
"hit_token_limit": false
|
| 166 |
},
|
|
|
|
| 172 |
"expected": 31.0,
|
| 173 |
"source": "openai/gsm8k",
|
| 174 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 175 |
+
"output": "<think>First find the total amount of ink he gets from the 25 pens: 25 pens * 5 pens/pens = 125 pens\nThen divide that number by the number of pens he can make from one set of five pens: 125 pens / 5 pens/pens = 25 pens</think>\nAnswer: 25",
|
| 176 |
"pass": false,
|
| 177 |
+
"hit_token_limit": false
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"id": "gsm8k-test-237",
|
|
|
|
| 184 |
"expected": 90.0,
|
| 185 |
"source": "openai/gsm8k",
|
| 186 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 187 |
+
"output": "<think>Johnny had a total of 20+10 = $30 to invest.\nThis amount tripled to 30*3 = $90.</think>\nAnswer: 90",
|
| 188 |
"pass": true,
|
| 189 |
"hit_token_limit": false
|
| 190 |
},
|
|
|
|
| 196 |
"expected": 68.0,
|
| 197 |
"source": "openai/gsm8k",
|
| 198 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 199 |
+
"output": "<think>Joan is 20 + 5 = 25 years old.\nMary is 25 - 2 = 23 years old.\nThe sum of their ages is 23 + 25 + 20 = 68 years.</think>\nAnswer: 68",
|
| 200 |
"pass": true,
|
| 201 |
"hit_token_limit": false
|
| 202 |
}
|
|
|
|
| 326 |
"expected": 75.0,
|
| 327 |
"source": "openai/gsm8k",
|
| 328 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 329 |
+
"output": "<think>There were 50/2 = 25 bugs in the garden.\nThe total number of insects in the garden is 25+50 = 75.</think>\nAnswer: 75",
|
| 330 |
"pass": true,
|
| 331 |
"hit_token_limit": false
|
| 332 |
},
|
|
|
|
| 338 |
"expected": 12.0,
|
| 339 |
"source": "openai/gsm8k",
|
| 340 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 341 |
+
"output": "<think>There are 36/3 = 12 penguins that jump in the ocean.\nThere are 36/3 = 12 penguins that go inside the cave.\n12 + 12 = 24 penguins are no longer sunbathing.\n36 - 24 = 12 penguins are still sunbathing.</think>\nAnswer: 12",
|
| 342 |
"pass": true,
|
| 343 |
"hit_token_limit": false
|
| 344 |
},
|
|
|
|
| 374 |
"expected": 31.0,
|
| 375 |
"source": "openai/gsm8k",
|
| 376 |
"source_revision": "740312add88f781978c0658806c59bc2815b9866",
|
| 377 |
+
"output": "<think>First find the total amount of ink he gets from the 25 pens: 25 pens * 5 pens/pens = 125 pens\nThen divide that number by the number of pens he can make from one set of empty pens: 125 pens / 5 pens/pens = 25 pens</think>\nAnswer: 25",
|
| 378 |
"pass": false,
|
| 379 |
"hit_token_limit": false
|
| 380 |
},
|
model-00001-of-00005.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3986791344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e4a9bfd682280e0271635d2b4329938f740373ce7d1fe3821d25b210bfbe45f
|
| 3 |
size 3986791344
|
model-00002-of-00005.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3926025416
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef0783b675748defc6ffc0a64580c578166ac499c5b83b64ceff8d5f60d12801
|
| 3 |
size 3926025416
|
model-00003-of-00005.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3926025440
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:30874d39316469ed6bff9cad62a9a576497dc47bc89886fbc67e8454f2733b09
|
| 3 |
size 3926025440
|
model-00004-of-00005.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3171040864
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6fa29d21bcb34fb555d70300cdd2fcb1932bbf2c4c5f2d7a985f0532a1601dce
|
| 3 |
size 3171040864
|
paired_bf16_outputs.json
ADDED
|
@@ -0,0 +1,1026 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base": [
|
| 3 |
+
{
|
| 4 |
+
"id": "heldout_00",
|
| 5 |
+
"prompt": "Return only the integer: 701 + 91 - 31.",
|
| 6 |
+
"expected": "761",
|
| 7 |
+
"kind": "integer",
|
| 8 |
+
"output": "761",
|
| 9 |
+
"pass": true
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"id": "heldout_01",
|
| 13 |
+
"prompt": "A crate has 704 red and 92 blue beads. Remove 32 red beads. Return only the total beads remaining.",
|
| 14 |
+
"expected": "764",
|
| 15 |
+
"kind": "integer",
|
| 16 |
+
"output": "The number of red beads left is 672. The number of blue beads is still 92, for a total of 764 beads.",
|
| 17 |
+
"pass": false
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"id": "heldout_02",
|
| 21 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_707, count=93.",
|
| 22 |
+
"expected": {
|
| 23 |
+
"item": "part_707",
|
| 24 |
+
"count": 93
|
| 25 |
+
},
|
| 26 |
+
"kind": "json",
|
| 27 |
+
"output": "{\"item\": \"part_707\", \"count\": 93}",
|
| 28 |
+
"pass": true
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"id": "heldout_03",
|
| 32 |
+
"prompt": "Return only one CSV value row, no header: name=Item710; quantity=94; code=K34.",
|
| 33 |
+
"expected": "Item710,94,K34",
|
| 34 |
+
"kind": "exact",
|
| 35 |
+
"output": "Item710,94,K34",
|
| 36 |
+
"pass": true
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"id": "heldout_04",
|
| 40 |
+
"prompt": "Copy exactly, no other text: Key_713-B95_C35",
|
| 41 |
+
"expected": "Key_713-B95_C35",
|
| 42 |
+
"kind": "exact",
|
| 43 |
+
"output": "Key_713-B95_C35",
|
| 44 |
+
"pass": true
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"id": "heldout_05",
|
| 48 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 96 + 36.",
|
| 49 |
+
"expected": "x * 96 + 36",
|
| 50 |
+
"kind": "code",
|
| 51 |
+
"output": "```python\ndef solve(x):\n return x * 96 + 36\n```",
|
| 52 |
+
"pass": true
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"id": "heldout_06",
|
| 56 |
+
"prompt": "Return only the integer: 719 + 97 - 37.",
|
| 57 |
+
"expected": "779",
|
| 58 |
+
"kind": "integer",
|
| 59 |
+
"output": "679",
|
| 60 |
+
"pass": false
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"id": "heldout_07",
|
| 64 |
+
"prompt": "A crate has 722 red and 98 blue beads. Remove 38 red beads. Return only the total beads remaining.",
|
| 65 |
+
"expected": "782",
|
| 66 |
+
"kind": "integer",
|
| 67 |
+
"output": "The number of red beads left is 722 - 38 = 684. The total number of beads left is 684 + 98 = 782.",
|
| 68 |
+
"pass": false
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"id": "heldout_08",
|
| 72 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_725, count=99.",
|
| 73 |
+
"expected": {
|
| 74 |
+
"item": "part_725",
|
| 75 |
+
"count": 99
|
| 76 |
+
},
|
| 77 |
+
"kind": "json",
|
| 78 |
+
"output": "{\"item\": \"part_725\", \"count\": 99}",
|
| 79 |
+
"pass": true
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"id": "heldout_09",
|
| 83 |
+
"prompt": "Return only one CSV value row, no header: name=Item728; quantity=100; code=K40.",
|
| 84 |
+
"expected": "Item728,100,K40",
|
| 85 |
+
"kind": "exact",
|
| 86 |
+
"output": "Item728,100,K40",
|
| 87 |
+
"pass": true
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"id": "heldout_10",
|
| 91 |
+
"prompt": "Copy exactly, no other text: Key_731-B101_C41",
|
| 92 |
+
"expected": "Key_731-B101_C41",
|
| 93 |
+
"kind": "exact",
|
| 94 |
+
"output": "Key_731-B101_C41",
|
| 95 |
+
"pass": true
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"id": "heldout_11",
|
| 99 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 102 + 42.",
|
| 100 |
+
"expected": "x * 102 + 42",
|
| 101 |
+
"kind": "code",
|
| 102 |
+
"output": "```python\ndef solve(x):\n return x * 102 + 42\n```",
|
| 103 |
+
"pass": true
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"id": "heldout_12",
|
| 107 |
+
"prompt": "Return only the integer: 737 + 103 - 43.",
|
| 108 |
+
"expected": "797",
|
| 109 |
+
"kind": "integer",
|
| 110 |
+
"output": "697",
|
| 111 |
+
"pass": false
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"id": "heldout_13",
|
| 115 |
+
"prompt": "A crate has 740 red and 104 blue beads. Remove 44 red beads. Return only the total beads remaining.",
|
| 116 |
+
"expected": "800",
|
| 117 |
+
"kind": "integer",
|
| 118 |
+
"output": "The number of red beads left is 740 - 44 = 696.\nThe total number of beads left is 696 + 104 = 800.",
|
| 119 |
+
"pass": false
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"id": "heldout_14",
|
| 123 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_743, count=105.",
|
| 124 |
+
"expected": {
|
| 125 |
+
"item": "part_743",
|
| 126 |
+
"count": 105
|
| 127 |
+
},
|
| 128 |
+
"kind": "json",
|
| 129 |
+
"output": "{\"item\": \"part_743\", \"count\": 105}",
|
| 130 |
+
"pass": true
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"id": "heldout_15",
|
| 134 |
+
"prompt": "Return only one CSV value row, no header: name=Item746; quantity=106; code=K46.",
|
| 135 |
+
"expected": "Item746,106,K46",
|
| 136 |
+
"kind": "exact",
|
| 137 |
+
"output": "Item746,106,K46",
|
| 138 |
+
"pass": true
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"id": "heldout_16",
|
| 142 |
+
"prompt": "Copy exactly, no other text: Key_749-B107_C47",
|
| 143 |
+
"expected": "Key_749-B107_C47",
|
| 144 |
+
"kind": "exact",
|
| 145 |
+
"output": "Key_749-B107_C47",
|
| 146 |
+
"pass": true
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"id": "heldout_17",
|
| 150 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 108 + 31.",
|
| 151 |
+
"expected": "x * 108 + 31",
|
| 152 |
+
"kind": "code",
|
| 153 |
+
"output": "```python\ndef solve(x):\n return x * 108 + 31\n```",
|
| 154 |
+
"pass": true
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"id": "heldout_18",
|
| 158 |
+
"prompt": "Return only the integer: 755 + 109 - 32.",
|
| 159 |
+
"expected": "832",
|
| 160 |
+
"kind": "integer",
|
| 161 |
+
"output": "732",
|
| 162 |
+
"pass": false
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"id": "heldout_19",
|
| 166 |
+
"prompt": "A crate has 758 red and 110 blue beads. Remove 33 red beads. Return only the total beads remaining.",
|
| 167 |
+
"expected": "835",
|
| 168 |
+
"kind": "integer",
|
| 169 |
+
"output": "The number of red beads left is 758 - 33 = 725. The total number of beads left is 725 + 110 = 835.",
|
| 170 |
+
"pass": false
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"id": "heldout_20",
|
| 174 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_761, count=111.",
|
| 175 |
+
"expected": {
|
| 176 |
+
"item": "part_761",
|
| 177 |
+
"count": 111
|
| 178 |
+
},
|
| 179 |
+
"kind": "json",
|
| 180 |
+
"output": "{\"item\": \"part_761\", \"count\": 111}",
|
| 181 |
+
"pass": true
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"id": "heldout_21",
|
| 185 |
+
"prompt": "Return only one CSV value row, no header: name=Item764; quantity=112; code=K35.",
|
| 186 |
+
"expected": "Item764,112,K35",
|
| 187 |
+
"kind": "exact",
|
| 188 |
+
"output": "Item764,112,K35",
|
| 189 |
+
"pass": true
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"id": "heldout_22",
|
| 193 |
+
"prompt": "Copy exactly, no other text: Key_767-B113_C36",
|
| 194 |
+
"expected": "Key_767-B113_C36",
|
| 195 |
+
"kind": "exact",
|
| 196 |
+
"output": "Key_767-B113_C36",
|
| 197 |
+
"pass": true
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"id": "heldout_23",
|
| 201 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 114 + 37.",
|
| 202 |
+
"expected": "x * 114 + 37",
|
| 203 |
+
"kind": "code",
|
| 204 |
+
"output": "```python\ndef solve(x):\n return x * 114 + 37\n```",
|
| 205 |
+
"pass": true
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"id": "heldout_24",
|
| 209 |
+
"prompt": "Return only the integer: 773 + 115 - 38.",
|
| 210 |
+
"expected": "850",
|
| 211 |
+
"kind": "integer",
|
| 212 |
+
"output": "750",
|
| 213 |
+
"pass": false
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"id": "heldout_25",
|
| 217 |
+
"prompt": "A crate has 776 red and 116 blue beads. Remove 39 red beads. Return only the total beads remaining.",
|
| 218 |
+
"expected": "853",
|
| 219 |
+
"kind": "integer",
|
| 220 |
+
"output": "The number of red beads left is 737 because 776 - 39 = 737. The total number of beads left is 737 + 116 = 853.",
|
| 221 |
+
"pass": false
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"id": "heldout_26",
|
| 225 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_779, count=117.",
|
| 226 |
+
"expected": {
|
| 227 |
+
"item": "part_779",
|
| 228 |
+
"count": 117
|
| 229 |
+
},
|
| 230 |
+
"kind": "json",
|
| 231 |
+
"output": "{\"item\": \"part_779\", \"count\": 117}",
|
| 232 |
+
"pass": true
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"id": "heldout_27",
|
| 236 |
+
"prompt": "Return only one CSV value row, no header: name=Item782; quantity=118; code=K41.",
|
| 237 |
+
"expected": "Item782,118,K41",
|
| 238 |
+
"kind": "exact",
|
| 239 |
+
"output": "Item782,118,K41",
|
| 240 |
+
"pass": true
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"id": "heldout_28",
|
| 244 |
+
"prompt": "Copy exactly, no other text: Key_785-B119_C42",
|
| 245 |
+
"expected": "Key_785-B119_C42",
|
| 246 |
+
"kind": "exact",
|
| 247 |
+
"output": "Key_785-B119_C42",
|
| 248 |
+
"pass": true
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"id": "heldout_29",
|
| 252 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 120 + 43.",
|
| 253 |
+
"expected": "x * 120 + 43",
|
| 254 |
+
"kind": "code",
|
| 255 |
+
"output": "```python\ndef solve(x):\n return x * 120 + 43\n```",
|
| 256 |
+
"pass": true
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"id": "heldout_30",
|
| 260 |
+
"prompt": "Return only the integer: 791 + 121 - 44.",
|
| 261 |
+
"expected": "868",
|
| 262 |
+
"kind": "integer",
|
| 263 |
+
"output": "768",
|
| 264 |
+
"pass": false
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"id": "heldout_31",
|
| 268 |
+
"prompt": "A crate has 794 red and 122 blue beads. Remove 45 red beads. Return only the total beads remaining.",
|
| 269 |
+
"expected": "871",
|
| 270 |
+
"kind": "integer",
|
| 271 |
+
"output": "The number of red beads left is 794 - 45 = 749. The total number of beads left is 749 + 122 = 871.",
|
| 272 |
+
"pass": false
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"id": "heldout_32",
|
| 276 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_797, count=123.",
|
| 277 |
+
"expected": {
|
| 278 |
+
"item": "part_797",
|
| 279 |
+
"count": 123
|
| 280 |
+
},
|
| 281 |
+
"kind": "json",
|
| 282 |
+
"output": "{\"item\": \"part_797\", \"count\": 123}",
|
| 283 |
+
"pass": true
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"id": "heldout_33",
|
| 287 |
+
"prompt": "Return only one CSV value row, no header: name=Item800; quantity=124; code=K47.",
|
| 288 |
+
"expected": "Item800,124,K47",
|
| 289 |
+
"kind": "exact",
|
| 290 |
+
"output": "Item800,124,K47",
|
| 291 |
+
"pass": true
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"id": "heldout_34",
|
| 295 |
+
"prompt": "Copy exactly, no other text: Key_803-B125_C31",
|
| 296 |
+
"expected": "Key_803-B125_C31",
|
| 297 |
+
"kind": "exact",
|
| 298 |
+
"output": "Key_803-B125_C31",
|
| 299 |
+
"pass": true
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"id": "heldout_35",
|
| 303 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 126 + 32.",
|
| 304 |
+
"expected": "x * 126 + 32",
|
| 305 |
+
"kind": "code",
|
| 306 |
+
"output": "```python\ndef solve(x):\n return x * 126 + 32\n```",
|
| 307 |
+
"pass": true
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"id": "heldout_36",
|
| 311 |
+
"prompt": "Return only the integer: 809 + 127 - 33.",
|
| 312 |
+
"expected": "903",
|
| 313 |
+
"kind": "integer",
|
| 314 |
+
"output": "803",
|
| 315 |
+
"pass": false
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"id": "heldout_37",
|
| 319 |
+
"prompt": "A crate has 812 red and 128 blue beads. Remove 34 red beads. Return only the total beads remaining.",
|
| 320 |
+
"expected": "906",
|
| 321 |
+
"kind": "integer",
|
| 322 |
+
"output": "The number of red beads left is 812 - 34 = 778. The total number of beads left is 778 + 128 = 906.",
|
| 323 |
+
"pass": false
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"id": "heldout_38",
|
| 327 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_815, count=129.",
|
| 328 |
+
"expected": {
|
| 329 |
+
"item": "part_815",
|
| 330 |
+
"count": 129
|
| 331 |
+
},
|
| 332 |
+
"kind": "json",
|
| 333 |
+
"output": "{\"item\": \"part_815\", \"count\": 129}",
|
| 334 |
+
"pass": true
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"id": "heldout_39",
|
| 338 |
+
"prompt": "Return only one CSV value row, no header: name=Item818; quantity=130; code=K36.",
|
| 339 |
+
"expected": "Item818,130,K36",
|
| 340 |
+
"kind": "exact",
|
| 341 |
+
"output": "Item818,130,K36",
|
| 342 |
+
"pass": true
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"id": "heldout_40",
|
| 346 |
+
"prompt": "Copy exactly, no other text: Key_821-B131_C37",
|
| 347 |
+
"expected": "Key_821-B131_C37",
|
| 348 |
+
"kind": "exact",
|
| 349 |
+
"output": "Key_821-B131_C37",
|
| 350 |
+
"pass": true
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"id": "heldout_41",
|
| 354 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 132 + 38.",
|
| 355 |
+
"expected": "x * 132 + 38",
|
| 356 |
+
"kind": "code",
|
| 357 |
+
"output": "```python\ndef solve(x):\n return x * 132 + 38\n```",
|
| 358 |
+
"pass": true
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"id": "heldout_42",
|
| 362 |
+
"prompt": "Return only the integer: 827 + 133 - 39.",
|
| 363 |
+
"expected": "921",
|
| 364 |
+
"kind": "integer",
|
| 365 |
+
"output": "841",
|
| 366 |
+
"pass": false
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"id": "heldout_43",
|
| 370 |
+
"prompt": "A crate has 830 red and 134 blue beads. Remove 40 red beads. Return only the total beads remaining.",
|
| 371 |
+
"expected": "924",
|
| 372 |
+
"kind": "integer",
|
| 373 |
+
"output": "The number of red beads left is 830 - 40 = 790. The total number of beads left is 790 + 134 = 924.",
|
| 374 |
+
"pass": false
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"id": "heldout_44",
|
| 378 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_833, count=135.",
|
| 379 |
+
"expected": {
|
| 380 |
+
"item": "part_833",
|
| 381 |
+
"count": 135
|
| 382 |
+
},
|
| 383 |
+
"kind": "json",
|
| 384 |
+
"output": "{\"item\": \"part_833\", \"count\": 135}",
|
| 385 |
+
"pass": true
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"id": "heldout_45",
|
| 389 |
+
"prompt": "Return only one CSV value row, no header: name=Item836; quantity=136; code=K42.",
|
| 390 |
+
"expected": "Item836,136,K42",
|
| 391 |
+
"kind": "exact",
|
| 392 |
+
"output": "Item836,136,K42",
|
| 393 |
+
"pass": true
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"id": "heldout_46",
|
| 397 |
+
"prompt": "Copy exactly, no other text: Key_839-B137_C43",
|
| 398 |
+
"expected": "Key_839-B137_C43",
|
| 399 |
+
"kind": "exact",
|
| 400 |
+
"output": "Key_839-B137_C43",
|
| 401 |
+
"pass": true
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"id": "heldout_47",
|
| 405 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 138 + 44.",
|
| 406 |
+
"expected": "x * 138 + 44",
|
| 407 |
+
"kind": "code",
|
| 408 |
+
"output": "```python\ndef solve(x):\n return x * 138 + 44\n```",
|
| 409 |
+
"pass": true
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"id": "heldout_48",
|
| 413 |
+
"prompt": "Return only the integer: 845 + 139 - 45.",
|
| 414 |
+
"expected": "939",
|
| 415 |
+
"kind": "integer",
|
| 416 |
+
"output": "839",
|
| 417 |
+
"pass": false
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"id": "heldout_49",
|
| 421 |
+
"prompt": "A crate has 848 red and 140 blue beads. Remove 46 red beads. Return only the total beads remaining.",
|
| 422 |
+
"expected": "942",
|
| 423 |
+
"kind": "integer",
|
| 424 |
+
"output": "The number of red beads left is 848 - 46 = 802.\nThe total number of beads left is 802 + 140 = 942.",
|
| 425 |
+
"pass": false
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"id": "heldout_50",
|
| 429 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_851, count=141.",
|
| 430 |
+
"expected": {
|
| 431 |
+
"item": "part_851",
|
| 432 |
+
"count": 141
|
| 433 |
+
},
|
| 434 |
+
"kind": "json",
|
| 435 |
+
"output": "{\"item\": \"part_851\", \"count\": 141}",
|
| 436 |
+
"pass": true
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"id": "heldout_51",
|
| 440 |
+
"prompt": "Return only one CSV value row, no header: name=Item854; quantity=142; code=K31.",
|
| 441 |
+
"expected": "Item854,142,K31",
|
| 442 |
+
"kind": "exact",
|
| 443 |
+
"output": "Item854,142,K31",
|
| 444 |
+
"pass": true
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"id": "heldout_52",
|
| 448 |
+
"prompt": "Copy exactly, no other text: Key_857-B143_C32",
|
| 449 |
+
"expected": "Key_857-B143_C32",
|
| 450 |
+
"kind": "exact",
|
| 451 |
+
"output": "Key_857-B143_C32",
|
| 452 |
+
"pass": true
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"id": "heldout_53",
|
| 456 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 144 + 33.",
|
| 457 |
+
"expected": "x * 144 + 33",
|
| 458 |
+
"kind": "code",
|
| 459 |
+
"output": "```python\ndef solve(x):\n return x * 144 + 33\n```",
|
| 460 |
+
"pass": true
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"id": "heldout_54",
|
| 464 |
+
"prompt": "Return only the integer: 863 + 145 - 34.",
|
| 465 |
+
"expected": "974",
|
| 466 |
+
"kind": "integer",
|
| 467 |
+
"output": "874",
|
| 468 |
+
"pass": false
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"id": "heldout_55",
|
| 472 |
+
"prompt": "A crate has 866 red and 146 blue beads. Remove 35 red beads. Return only the total beads remaining.",
|
| 473 |
+
"expected": "977",
|
| 474 |
+
"kind": "integer",
|
| 475 |
+
"output": "The number of red beads left is 866 - 35 = 831. The total number of beads left is 831 + 146 = 977.",
|
| 476 |
+
"pass": false
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"id": "heldout_56",
|
| 480 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_869, count=147.",
|
| 481 |
+
"expected": {
|
| 482 |
+
"item": "part_869",
|
| 483 |
+
"count": 147
|
| 484 |
+
},
|
| 485 |
+
"kind": "json",
|
| 486 |
+
"output": "{\"item\": \"part_869\", \"count\": 147}",
|
| 487 |
+
"pass": true
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"id": "heldout_57",
|
| 491 |
+
"prompt": "Return only one CSV value row, no header: name=Item872; quantity=148; code=K37.",
|
| 492 |
+
"expected": "Item872,148,K37",
|
| 493 |
+
"kind": "exact",
|
| 494 |
+
"output": "Item872,148,K37",
|
| 495 |
+
"pass": true
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"id": "heldout_58",
|
| 499 |
+
"prompt": "Copy exactly, no other text: Key_875-B149_C38",
|
| 500 |
+
"expected": "Key_875-B149_C38",
|
| 501 |
+
"kind": "exact",
|
| 502 |
+
"output": "Key_875-B149_C38",
|
| 503 |
+
"pass": true
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"id": "heldout_59",
|
| 507 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 150 + 39.",
|
| 508 |
+
"expected": "x * 150 + 39",
|
| 509 |
+
"kind": "code",
|
| 510 |
+
"output": "```python\ndef solve(x):\n return x * 150 + 39\n```",
|
| 511 |
+
"pass": true
|
| 512 |
+
}
|
| 513 |
+
],
|
| 514 |
+
"astheral": [
|
| 515 |
+
{
|
| 516 |
+
"id": "heldout_00",
|
| 517 |
+
"prompt": "Return only the integer: 701 + 91 - 31.",
|
| 518 |
+
"expected": "761",
|
| 519 |
+
"kind": "integer",
|
| 520 |
+
"output": "761",
|
| 521 |
+
"pass": true
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"id": "heldout_01",
|
| 525 |
+
"prompt": "A crate has 704 red and 92 blue beads. Remove 32 red beads. Return only the total beads remaining.",
|
| 526 |
+
"expected": "764",
|
| 527 |
+
"kind": "integer",
|
| 528 |
+
"output": "The number of red beads left is 672. The number of blue beads is still 92, for a total of 764 beads.",
|
| 529 |
+
"pass": false
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"id": "heldout_02",
|
| 533 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_707, count=93.",
|
| 534 |
+
"expected": {
|
| 535 |
+
"item": "part_707",
|
| 536 |
+
"count": 93
|
| 537 |
+
},
|
| 538 |
+
"kind": "json",
|
| 539 |
+
"output": "{\"item\": \"part_707\", \"count\": 93}",
|
| 540 |
+
"pass": true
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"id": "heldout_03",
|
| 544 |
+
"prompt": "Return only one CSV value row, no header: name=Item710; quantity=94; code=K34.",
|
| 545 |
+
"expected": "Item710,94,K34",
|
| 546 |
+
"kind": "exact",
|
| 547 |
+
"output": "Item710,94,K34",
|
| 548 |
+
"pass": true
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"id": "heldout_04",
|
| 552 |
+
"prompt": "Copy exactly, no other text: Key_713-B95_C35",
|
| 553 |
+
"expected": "Key_713-B95_C35",
|
| 554 |
+
"kind": "exact",
|
| 555 |
+
"output": "Key_713-B95_C35",
|
| 556 |
+
"pass": true
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"id": "heldout_05",
|
| 560 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 96 + 36.",
|
| 561 |
+
"expected": "x * 96 + 36",
|
| 562 |
+
"kind": "code",
|
| 563 |
+
"output": "```python\ndef solve(x):\n return x * 96 + 36\n```",
|
| 564 |
+
"pass": true
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"id": "heldout_06",
|
| 568 |
+
"prompt": "Return only the integer: 719 + 97 - 37.",
|
| 569 |
+
"expected": "779",
|
| 570 |
+
"kind": "integer",
|
| 571 |
+
"output": "779",
|
| 572 |
+
"pass": true
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"id": "heldout_07",
|
| 576 |
+
"prompt": "A crate has 722 red and 98 blue beads. Remove 38 red beads. Return only the total beads remaining.",
|
| 577 |
+
"expected": "782",
|
| 578 |
+
"kind": "integer",
|
| 579 |
+
"output": "The number of red beads left is 684. The number of beads left is 684 + 98 = 782.",
|
| 580 |
+
"pass": false
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"id": "heldout_08",
|
| 584 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_725, count=99.",
|
| 585 |
+
"expected": {
|
| 586 |
+
"item": "part_725",
|
| 587 |
+
"count": 99
|
| 588 |
+
},
|
| 589 |
+
"kind": "json",
|
| 590 |
+
"output": "{\"item\": \"part_725\", \"count\": 99}",
|
| 591 |
+
"pass": true
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"id": "heldout_09",
|
| 595 |
+
"prompt": "Return only one CSV value row, no header: name=Item728; quantity=100; code=K40.",
|
| 596 |
+
"expected": "Item728,100,K40",
|
| 597 |
+
"kind": "exact",
|
| 598 |
+
"output": "Item728,100,K40",
|
| 599 |
+
"pass": true
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"id": "heldout_10",
|
| 603 |
+
"prompt": "Copy exactly, no other text: Key_731-B101_C41",
|
| 604 |
+
"expected": "Key_731-B101_C41",
|
| 605 |
+
"kind": "exact",
|
| 606 |
+
"output": "Key_731-B101_C41",
|
| 607 |
+
"pass": true
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"id": "heldout_11",
|
| 611 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 102 + 42.",
|
| 612 |
+
"expected": "x * 102 + 42",
|
| 613 |
+
"kind": "code",
|
| 614 |
+
"output": "```python\ndef solve(x):\n return x * 102 + 42\n```",
|
| 615 |
+
"pass": true
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"id": "heldout_12",
|
| 619 |
+
"prompt": "Return only the integer: 737 + 103 - 43.",
|
| 620 |
+
"expected": "797",
|
| 621 |
+
"kind": "integer",
|
| 622 |
+
"output": "797",
|
| 623 |
+
"pass": true
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"id": "heldout_13",
|
| 627 |
+
"prompt": "A crate has 740 red and 104 blue beads. Remove 44 red beads. Return only the total beads remaining.",
|
| 628 |
+
"expected": "800",
|
| 629 |
+
"kind": "integer",
|
| 630 |
+
"output": "The number of red beads left is 696. The number of blue beads is unchanged at 104, for a total of 800 beads remaining.",
|
| 631 |
+
"pass": false
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"id": "heldout_14",
|
| 635 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_743, count=105.",
|
| 636 |
+
"expected": {
|
| 637 |
+
"item": "part_743",
|
| 638 |
+
"count": 105
|
| 639 |
+
},
|
| 640 |
+
"kind": "json",
|
| 641 |
+
"output": "{\"item\": \"part_743\", \"count\": 105}",
|
| 642 |
+
"pass": true
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"id": "heldout_15",
|
| 646 |
+
"prompt": "Return only one CSV value row, no header: name=Item746; quantity=106; code=K46.",
|
| 647 |
+
"expected": "Item746,106,K46",
|
| 648 |
+
"kind": "exact",
|
| 649 |
+
"output": "Item746,106,K46",
|
| 650 |
+
"pass": true
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"id": "heldout_16",
|
| 654 |
+
"prompt": "Copy exactly, no other text: Key_749-B107_C47",
|
| 655 |
+
"expected": "Key_749-B107_C47",
|
| 656 |
+
"kind": "exact",
|
| 657 |
+
"output": "Key_749-B107_C47",
|
| 658 |
+
"pass": true
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"id": "heldout_17",
|
| 662 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 108 + 31.",
|
| 663 |
+
"expected": "x * 108 + 31",
|
| 664 |
+
"kind": "code",
|
| 665 |
+
"output": "```python\ndef solve(x):\n return x * 108 + 31\n```",
|
| 666 |
+
"pass": true
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"id": "heldout_18",
|
| 670 |
+
"prompt": "Return only the integer: 755 + 109 - 32.",
|
| 671 |
+
"expected": "832",
|
| 672 |
+
"kind": "integer",
|
| 673 |
+
"output": "732",
|
| 674 |
+
"pass": false
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"id": "heldout_19",
|
| 678 |
+
"prompt": "A crate has 758 red and 110 blue beads. Remove 33 red beads. Return only the total beads remaining.",
|
| 679 |
+
"expected": "835",
|
| 680 |
+
"kind": "integer",
|
| 681 |
+
"output": "The number of red beads left is 758 - 33 = 725. The total number of beads left is 725 + 110 = 835.",
|
| 682 |
+
"pass": false
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"id": "heldout_20",
|
| 686 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_761, count=111.",
|
| 687 |
+
"expected": {
|
| 688 |
+
"item": "part_761",
|
| 689 |
+
"count": 111
|
| 690 |
+
},
|
| 691 |
+
"kind": "json",
|
| 692 |
+
"output": "{\"item\": \"part_761\", \"count\": 111}",
|
| 693 |
+
"pass": true
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"id": "heldout_21",
|
| 697 |
+
"prompt": "Return only one CSV value row, no header: name=Item764; quantity=112; code=K35.",
|
| 698 |
+
"expected": "Item764,112,K35",
|
| 699 |
+
"kind": "exact",
|
| 700 |
+
"output": "Item764,112,K35",
|
| 701 |
+
"pass": true
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"id": "heldout_22",
|
| 705 |
+
"prompt": "Copy exactly, no other text: Key_767-B113_C36",
|
| 706 |
+
"expected": "Key_767-B113_C36",
|
| 707 |
+
"kind": "exact",
|
| 708 |
+
"output": "Key_767-B113_C36",
|
| 709 |
+
"pass": true
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"id": "heldout_23",
|
| 713 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 114 + 37.",
|
| 714 |
+
"expected": "x * 114 + 37",
|
| 715 |
+
"kind": "code",
|
| 716 |
+
"output": "```python\ndef solve(x):\n return x * 114 + 37\n```",
|
| 717 |
+
"pass": true
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"id": "heldout_24",
|
| 721 |
+
"prompt": "Return only the integer: 773 + 115 - 38.",
|
| 722 |
+
"expected": "850",
|
| 723 |
+
"kind": "integer",
|
| 724 |
+
"output": "750",
|
| 725 |
+
"pass": false
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"id": "heldout_25",
|
| 729 |
+
"prompt": "A crate has 776 red and 116 blue beads. Remove 39 red beads. Return only the total beads remaining.",
|
| 730 |
+
"expected": "853",
|
| 731 |
+
"kind": "integer",
|
| 732 |
+
"output": "The number of red beads left is 737. There are 116 blue beads, 737 red beads, and 1533 beads left.",
|
| 733 |
+
"pass": false
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"id": "heldout_26",
|
| 737 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_779, count=117.",
|
| 738 |
+
"expected": {
|
| 739 |
+
"item": "part_779",
|
| 740 |
+
"count": 117
|
| 741 |
+
},
|
| 742 |
+
"kind": "json",
|
| 743 |
+
"output": "{\"item\": \"part_779\", \"count\": 117}",
|
| 744 |
+
"pass": true
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"id": "heldout_27",
|
| 748 |
+
"prompt": "Return only one CSV value row, no header: name=Item782; quantity=118; code=K41.",
|
| 749 |
+
"expected": "Item782,118,K41",
|
| 750 |
+
"kind": "exact",
|
| 751 |
+
"output": "Item782,118,K41",
|
| 752 |
+
"pass": true
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"id": "heldout_28",
|
| 756 |
+
"prompt": "Copy exactly, no other text: Key_785-B119_C42",
|
| 757 |
+
"expected": "Key_785-B119_C42",
|
| 758 |
+
"kind": "exact",
|
| 759 |
+
"output": "Key_785-B119_C42",
|
| 760 |
+
"pass": true
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"id": "heldout_29",
|
| 764 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 120 + 43.",
|
| 765 |
+
"expected": "x * 120 + 43",
|
| 766 |
+
"kind": "code",
|
| 767 |
+
"output": "```python\ndef solve(x):\n return x * 120 + 43\n```",
|
| 768 |
+
"pass": true
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"id": "heldout_30",
|
| 772 |
+
"prompt": "Return only the integer: 791 + 121 - 44.",
|
| 773 |
+
"expected": "868",
|
| 774 |
+
"kind": "integer",
|
| 775 |
+
"output": "768",
|
| 776 |
+
"pass": false
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"id": "heldout_31",
|
| 780 |
+
"prompt": "A crate has 794 red and 122 blue beads. Remove 45 red beads. Return only the total beads remaining.",
|
| 781 |
+
"expected": "871",
|
| 782 |
+
"kind": "integer",
|
| 783 |
+
"output": "The number of red beads left is 749 because 794 - 45 = 749. The total number of beads left is 749 + 122 = 871.",
|
| 784 |
+
"pass": false
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"id": "heldout_32",
|
| 788 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_797, count=123.",
|
| 789 |
+
"expected": {
|
| 790 |
+
"item": "part_797",
|
| 791 |
+
"count": 123
|
| 792 |
+
},
|
| 793 |
+
"kind": "json",
|
| 794 |
+
"output": "{\"item\": \"part_797\", \"count\": 123}",
|
| 795 |
+
"pass": true
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"id": "heldout_33",
|
| 799 |
+
"prompt": "Return only one CSV value row, no header: name=Item800; quantity=124; code=K47.",
|
| 800 |
+
"expected": "Item800,124,K47",
|
| 801 |
+
"kind": "exact",
|
| 802 |
+
"output": "Item800,124,K47",
|
| 803 |
+
"pass": true
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"id": "heldout_34",
|
| 807 |
+
"prompt": "Copy exactly, no other text: Key_803-B125_C31",
|
| 808 |
+
"expected": "Key_803-B125_C31",
|
| 809 |
+
"kind": "exact",
|
| 810 |
+
"output": "Key_803-B125_C31",
|
| 811 |
+
"pass": true
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"id": "heldout_35",
|
| 815 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 126 + 32.",
|
| 816 |
+
"expected": "x * 126 + 32",
|
| 817 |
+
"kind": "code",
|
| 818 |
+
"output": "```python\ndef solve(x):\n return x * 126 + 32\n```",
|
| 819 |
+
"pass": true
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"id": "heldout_36",
|
| 823 |
+
"prompt": "Return only the integer: 809 + 127 - 33.",
|
| 824 |
+
"expected": "903",
|
| 825 |
+
"kind": "integer",
|
| 826 |
+
"output": "803",
|
| 827 |
+
"pass": false
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"id": "heldout_37",
|
| 831 |
+
"prompt": "A crate has 812 red and 128 blue beads. Remove 34 red beads. Return only the total beads remaining.",
|
| 832 |
+
"expected": "906",
|
| 833 |
+
"kind": "integer",
|
| 834 |
+
"output": "The number of red beads left is 812 - 34 = 778. The total number of beads left is 778 + 128 = 906.",
|
| 835 |
+
"pass": false
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"id": "heldout_38",
|
| 839 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_815, count=129.",
|
| 840 |
+
"expected": {
|
| 841 |
+
"item": "part_815",
|
| 842 |
+
"count": 129
|
| 843 |
+
},
|
| 844 |
+
"kind": "json",
|
| 845 |
+
"output": "{\"item\": \"part_815\", \"count\": 129}",
|
| 846 |
+
"pass": true
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"id": "heldout_39",
|
| 850 |
+
"prompt": "Return only one CSV value row, no header: name=Item818; quantity=130; code=K36.",
|
| 851 |
+
"expected": "Item818,130,K36",
|
| 852 |
+
"kind": "exact",
|
| 853 |
+
"output": "Item818,130,K36",
|
| 854 |
+
"pass": true
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"id": "heldout_40",
|
| 858 |
+
"prompt": "Copy exactly, no other text: Key_821-B131_C37",
|
| 859 |
+
"expected": "Key_821-B131_C37",
|
| 860 |
+
"kind": "exact",
|
| 861 |
+
"output": "Key_821-B131_C37",
|
| 862 |
+
"pass": true
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"id": "heldout_41",
|
| 866 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 132 + 38.",
|
| 867 |
+
"expected": "x * 132 + 38",
|
| 868 |
+
"kind": "code",
|
| 869 |
+
"output": "```python\ndef solve(x):\n return x * 132 + 38\n```",
|
| 870 |
+
"pass": true
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"id": "heldout_42",
|
| 874 |
+
"prompt": "Return only the integer: 827 + 133 - 39.",
|
| 875 |
+
"expected": "921",
|
| 876 |
+
"kind": "integer",
|
| 877 |
+
"output": "821",
|
| 878 |
+
"pass": false
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"id": "heldout_43",
|
| 882 |
+
"prompt": "A crate has 830 red and 134 blue beads. Remove 40 red beads. Return only the total beads remaining.",
|
| 883 |
+
"expected": "924",
|
| 884 |
+
"kind": "integer",
|
| 885 |
+
"output": "The number of red beads left is 790 because 830 - 40 = 790. The total number of beads left is 790 + 134 = 924.",
|
| 886 |
+
"pass": false
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"id": "heldout_44",
|
| 890 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_833, count=135.",
|
| 891 |
+
"expected": {
|
| 892 |
+
"item": "part_833",
|
| 893 |
+
"count": 135
|
| 894 |
+
},
|
| 895 |
+
"kind": "json",
|
| 896 |
+
"output": "{\"item\": \"part_833\", \"count\": 135}",
|
| 897 |
+
"pass": true
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"id": "heldout_45",
|
| 901 |
+
"prompt": "Return only one CSV value row, no header: name=Item836; quantity=136; code=K42.",
|
| 902 |
+
"expected": "Item836,136,K42",
|
| 903 |
+
"kind": "exact",
|
| 904 |
+
"output": "Item836,136,K42",
|
| 905 |
+
"pass": true
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"id": "heldout_46",
|
| 909 |
+
"prompt": "Copy exactly, no other text: Key_839-B137_C43",
|
| 910 |
+
"expected": "Key_839-B137_C43",
|
| 911 |
+
"kind": "exact",
|
| 912 |
+
"output": "Key_839-B137_C43",
|
| 913 |
+
"pass": true
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"id": "heldout_47",
|
| 917 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 138 + 44.",
|
| 918 |
+
"expected": "x * 138 + 44",
|
| 919 |
+
"kind": "code",
|
| 920 |
+
"output": "```python\ndef solve(x):\n return x * 138 + 44\n```",
|
| 921 |
+
"pass": true
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"id": "heldout_48",
|
| 925 |
+
"prompt": "Return only the integer: 845 + 139 - 45.",
|
| 926 |
+
"expected": "939",
|
| 927 |
+
"kind": "integer",
|
| 928 |
+
"output": "839",
|
| 929 |
+
"pass": false
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"id": "heldout_49",
|
| 933 |
+
"prompt": "A crate has 848 red and 140 blue beads. Remove 46 red beads. Return only the total beads remaining.",
|
| 934 |
+
"expected": "942",
|
| 935 |
+
"kind": "integer",
|
| 936 |
+
"output": "The number of red beads left is 848 - 46 = 802. The total number of beads left is 802 + 140 = 942.",
|
| 937 |
+
"pass": false
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"id": "heldout_50",
|
| 941 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_851, count=141.",
|
| 942 |
+
"expected": {
|
| 943 |
+
"item": "part_851",
|
| 944 |
+
"count": 141
|
| 945 |
+
},
|
| 946 |
+
"kind": "json",
|
| 947 |
+
"output": "{\"item\": \"part_851\", \"count\": 141}",
|
| 948 |
+
"pass": true
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"id": "heldout_51",
|
| 952 |
+
"prompt": "Return only one CSV value row, no header: name=Item854; quantity=142; code=K31.",
|
| 953 |
+
"expected": "Item854,142,K31",
|
| 954 |
+
"kind": "exact",
|
| 955 |
+
"output": "Item854,142,K31",
|
| 956 |
+
"pass": true
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"id": "heldout_52",
|
| 960 |
+
"prompt": "Copy exactly, no other text: Key_857-B143_C32",
|
| 961 |
+
"expected": "Key_857-B143_C32",
|
| 962 |
+
"kind": "exact",
|
| 963 |
+
"output": "Key_857-B143_C32",
|
| 964 |
+
"pass": true
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"id": "heldout_53",
|
| 968 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 144 + 33.",
|
| 969 |
+
"expected": "x * 144 + 33",
|
| 970 |
+
"kind": "code",
|
| 971 |
+
"output": "```python\ndef solve(x):\n return x * 144 + 33\n```",
|
| 972 |
+
"pass": true
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"id": "heldout_54",
|
| 976 |
+
"prompt": "Return only the integer: 863 + 145 - 34.",
|
| 977 |
+
"expected": "974",
|
| 978 |
+
"kind": "integer",
|
| 979 |
+
"output": "874",
|
| 980 |
+
"pass": false
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"id": "heldout_55",
|
| 984 |
+
"prompt": "A crate has 866 red and 146 blue beads. Remove 35 red beads. Return only the total beads remaining.",
|
| 985 |
+
"expected": "977",
|
| 986 |
+
"kind": "integer",
|
| 987 |
+
"output": "The number of red beads left is 831. The number of blue beads is still 146, for a total of 977 beads.",
|
| 988 |
+
"pass": false
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"id": "heldout_56",
|
| 992 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_869, count=147.",
|
| 993 |
+
"expected": {
|
| 994 |
+
"item": "part_869",
|
| 995 |
+
"count": 147
|
| 996 |
+
},
|
| 997 |
+
"kind": "json",
|
| 998 |
+
"output": "{\"item\": \"part_869\", \"count\": 147}",
|
| 999 |
+
"pass": true
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"id": "heldout_57",
|
| 1003 |
+
"prompt": "Return only one CSV value row, no header: name=Item872; quantity=148; code=K37.",
|
| 1004 |
+
"expected": "Item872,148,K37",
|
| 1005 |
+
"kind": "exact",
|
| 1006 |
+
"output": "Item872,148,K37",
|
| 1007 |
+
"pass": true
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"id": "heldout_58",
|
| 1011 |
+
"prompt": "Copy exactly, no other text: Key_875-B149_C38",
|
| 1012 |
+
"expected": "Key_875-B149_C38",
|
| 1013 |
+
"kind": "exact",
|
| 1014 |
+
"output": "Key_875-B149_C38",
|
| 1015 |
+
"pass": true
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"id": "heldout_59",
|
| 1019 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 150 + 39.",
|
| 1020 |
+
"expected": "x * 150 + 39",
|
| 1021 |
+
"kind": "code",
|
| 1022 |
+
"output": "```python\ndef solve(x):\n return x * 150 + 39\n```",
|
| 1023 |
+
"pass": true
|
| 1024 |
+
}
|
| 1025 |
+
]
|
| 1026 |
+
}
|
train.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
train.py
CHANGED
|
@@ -1,173 +1,128 @@
|
|
| 1 |
-
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
| 3 |
from pathlib import Path
|
| 4 |
import torch
|
| 5 |
-
from
|
| 6 |
-
from
|
| 7 |
-
from
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
c("confirm_time_2","How many minutes are in two hours and seventeen minutes? Explain briefly.",137,"number","thinking")
|
| 35 |
-
for fn in ["pack_adjacent","compress_chars"]:
|
| 36 |
-
confirm.append(dict(id="confirm_code_"+fn,prompt=f"Write only Python code for {fn}(text), returning consecutive (character, count) tuples as a list.",kind="code",expected=None,mode="instruct",group="confirmation",tests=f"assert {fn}('aabbba')==[('a',2),('b',3),('a',1)]\nassert {fn}('')==[]\nassert {fn}('x')==[('x',1)]"))
|
| 37 |
-
ALL=CASES+confirm
|
| 38 |
-
MAX_TOKENS=512
|
| 39 |
-
class Samples(Dataset):
|
| 40 |
-
def __init__(self,rows):
|
| 41 |
-
self.items=[];self.kept=[];self.skipped=0
|
| 42 |
-
forbidden={x["prompt"] for x in ALL}
|
| 43 |
-
for row in rows:
|
| 44 |
-
assert not any(m["role"]=="user" and m["content"] in forbidden for m in row["messages"])
|
| 45 |
-
prefix=tok.apply_chat_template(row["messages"][:-1],tokenize=True,add_generation_prompt=True)
|
| 46 |
-
full=tok.apply_chat_template(row["messages"],tokenize=True,add_generation_prompt=False)
|
| 47 |
-
assert full[:len(prefix)]==prefix
|
| 48 |
-
if len(full)>MAX_TOKENS:self.skipped+=1;continue
|
| 49 |
-
self.items.append({"input_ids":full,"labels":[-100]*len(prefix)+full[len(prefix):]});self.kept.append(row)
|
| 50 |
-
def __len__(self):return len(self.items)
|
| 51 |
-
def __getitem__(self,index):return self.items[index]
|
| 52 |
-
def collate(batch):
|
| 53 |
-
n=max(len(r["input_ids"]) for r in batch)
|
| 54 |
-
return {k:torch.tensor([r[k]+[fill]*(n-len(r[k])) for r in batch]) for k,fill in [("input_ids",128004),("labels",-100)]}|{"attention_mask":torch.tensor([[1]*len(r["input_ids"])+[0]*(n-len(r["input_ids"])) for r in batch])}
|
| 55 |
-
train_records=[json.loads(s) for s in (ROOT/"repair_train.jsonl").read_text().splitlines()]
|
| 56 |
-
valid_records=[json.loads(s) for s in (ROOT/"repair_validation.jsonl").read_text().splitlines()]
|
| 57 |
-
train=Samples(train_records)
|
| 58 |
-
valid=Samples(valid_records)
|
| 59 |
-
assert len(train)>1200 and len(valid)>40
|
| 60 |
-
print("BF16_DATA",len(train),len(valid),"skipped",train.skipped,flush=True)
|
| 61 |
-
baseline=json.loads(read("v2-baseline.json"))["outputs"]
|
| 62 |
-
for row in baseline:row["pass"]=check(row,row["output"])
|
| 63 |
-
(OUT/"base_outputs.json").write_text(json.dumps(baseline,indent=2))
|
| 64 |
-
base=AutoModelForCausalLM.from_pretrained(BASE,revision=REV,torch_dtype=torch.bfloat16,device_map={"":"cuda:0"},attn_implementation="sdpa")
|
| 65 |
-
adapter_root=snapshot_download(REPO,revision=SOURCE_REV,allow_patterns=["repair-results/adapter/*"],local_dir="/tmp/initial-adapter")
|
| 66 |
-
model=PeftModel.from_pretrained(base,Path(adapter_root)/"repair-results/adapter",is_trainable=True)
|
| 67 |
-
model.config.use_cache=False;model.enable_input_require_grads()
|
| 68 |
-
model.gradient_checkpointing_enable(gradient_checkpointing_kwargs={"use_reentrant":False})
|
| 69 |
-
trainable=sum(p.numel() for p in model.parameters() if p.requires_grad);assert 0<trainable<100000000
|
| 70 |
-
# Verify actual BF16 backward memory before the optimizer run.
|
| 71 |
-
for limit in [512,384]:
|
| 72 |
-
MAX_TOKENS=limit
|
| 73 |
-
train=Samples(train_records);valid=Samples(valid_records)
|
| 74 |
-
largest=max(train.items,key=lambda r:len(r["input_ids"]))
|
| 75 |
-
probe={k:v.to("cuda") for k,v in collate([largest]).items()}
|
| 76 |
try:
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
if
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
@torch.inference_mode()
|
| 95 |
-
def
|
| 96 |
-
model.eval();
|
| 97 |
-
for
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
|
| 128 |
-
(
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
(OUT/"confirmation_outputs.json").write_text(json.dumps(merged_confirm,indent=2))
|
| 147 |
-
(OUT/"merged_regressions_outputs.json").write_text(json.dumps([r for r in merged if r["group"]=="regression"],indent=2))
|
| 148 |
-
metrics["merged_scores"]=scores(merged);metrics["quality_gate_passed"]=gate(merged);metrics["elapsed_seconds"]=time.monotonic()-start
|
| 149 |
-
(OUT/"evaluation.json").write_text(json.dumps(metrics,indent=2))
|
| 150 |
-
api.upload_folder(repo_id=REPO,folder_path=OUT,path_in_repo="repair2-results",allow_patterns=["evaluation.json","astheral_merged_outputs.json","astheral_outputs.json","merged_regressions_outputs.json","confirmation_outputs.json","trainer_log.json"],commit_message="Record complete reloaded second BF16 repair results")
|
| 151 |
-
print("REPAIR_GATE",metrics["quality_gate_passed"],json.dumps(metrics["merged_scores"]),flush=True)
|
| 152 |
-
if not metrics["quality_gate_passed"]:raise SystemExit(3)
|
| 153 |
-
for n in ["LICENSE","UPSTREAM_LLAMA3_LICENSE","NOTICE"]:shutil.copy2(Path(hf_hub_download(REPO,n,revision=SOURCE_REV)),OUT/n)
|
| 154 |
-
for n in ["DATA_LICENSE_GSM8K.txt","DATA_LICENSE_APACHE2.txt"]:shutil.copy2(Path(hf_hub_download(REPO,"v2/"+n,revision=SOURCE_REV)),OUT/n)
|
| 155 |
-
(OUT/"DATA_LICENSE_ODCBY.txt").write_text("Contains data from bigcode/self-oss-instruct-sc2-exec-filter-50k under ODC Attribution v1.0: https://opendatacommons.org/licenses/by/1-0/.\n")
|
| 156 |
-
(OUT/"NOTICE").write_text((OUT/"NOTICE").read_text()+"\nTraining sources: HuggingFaceTB/SmolTalk (Apache 2.0), openai/gsm8k (MIT; Copyright (c) 2021 OpenAI), bigcode/self-oss-instruct-sc2-exec-filter-50k (ODC Attribution), and original Astheral repair examples (CC0-1.0).\n")
|
| 157 |
-
sources=json.loads(read("v2/data_sources.json"));sources["repair"]=provenance
|
| 158 |
-
(OUT/"data_sources.json").write_text(json.dumps(sources,indent=2))
|
| 159 |
-
(OUT/"DATA.md").write_text("# Training data\n\nOriginal and repair training examples are included. data_sources.json records provenance and component licenses. Original Astheral examples are CC0-1.0; third-party examples retain their source terms. Training stages and exact optimizer steps are recorded in evaluation.json.\n")
|
| 160 |
-
(OUT/"reload_smoke_test.json").write_text(json.dumps({"loaded_from":"saved merged BF16 checkpoint","core_checks_passed":38,"confirmation_checks_passed":len(confirm)},indent=2))
|
| 161 |
-
for n in ["inference.py","requirements.txt","Dockerfile"]:shutil.copy2(ROOT/n,OUT/n)
|
| 162 |
-
(OUT/"README.md").write_text((ROOT/"MODEL_CARD.md").read_text()+"\n## Verified core results\n\n"+json.dumps(metrics["merged_scores"],indent=2)+"\n")
|
| 163 |
-
manifest={}
|
| 164 |
-
for f in OUT.rglob("*"):
|
| 165 |
-
if f.is_file():
|
| 166 |
-
h=hashlib.sha256()
|
| 167 |
-
with f.open("rb") as handle:
|
| 168 |
-
for part in iter(lambda:handle.read(8388608),b""):h.update(part)
|
| 169 |
-
manifest[str(f.relative_to(OUT))]={"bytes":f.stat().st_size,"sha256":h.hexdigest()}
|
| 170 |
-
(OUT/"artifact_manifest.json").write_text(json.dumps(manifest,indent=2))
|
| 171 |
-
api.create_tag(REPO,tag="v1",revision=api.model_info(REPO).sha,exist_ok=True)
|
| 172 |
-
api.upload_folder(repo_id=REPO,folder_path=OUT,delete_patterns=["training/*","coding_smoke_test.json"],commit_message="Release Astheral v2 after native-BF16 repair and all core/confirmation checks")
|
| 173 |
-
print("REPAIR_PUBLISHED",api.model_info(REPO).sha,flush=True)
|
|
|
|
| 1 |
+
# /// script
|
| 2 |
+
# dependencies = ["torch==2.8.0", "transformers==4.56.2", "peft==0.17.1", "accelerate==1.10.1", "huggingface_hub==0.34.4", "bitsandbytes==0.47.0", "safetensors==0.6.2"]
|
| 3 |
+
# ///
|
| 4 |
+
"""Budget-bounded V3 training with paired, disjoint evaluation and reload gate."""
|
| 5 |
+
import os, json, random, time, re, gc, hashlib, traceback
|
| 6 |
from pathlib import Path
|
| 7 |
import torch
|
| 8 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, Trainer, TrainingArguments, TrainerCallback, set_seed
|
| 9 |
+
from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training, PeftModel
|
| 10 |
+
from huggingface_hub import HfApi, hf_hub_download
|
| 11 |
+
|
| 12 |
+
REPO='SEE7/Astheral'; REV='d9aa899722be4788f57107c2951c72d0ec6614ee'
|
| 13 |
+
OUT=Path('/tmp/astheral-v3'); OUT.mkdir(exist_ok=True)
|
| 14 |
+
api=HfApi(); started=time.monotonic(); set_seed(20261113); rng=random.Random(20261113)
|
| 15 |
+
assert torch.cuda.is_available() and 'A10G' in torch.cuda.get_device_name(0)
|
| 16 |
+
branch='v3-candidate-20261113'
|
| 17 |
+
api.create_branch(REPO,branch=branch,revision=REV,exist_ok=True)
|
| 18 |
+
settings=json.loads(Path(hf_hub_download(REPO,'settings.json',revision=REV)).read_text())
|
| 19 |
+
tok=AutoTokenizer.from_pretrained(REPO,revision=REV); tok.pad_token_id=128004
|
| 20 |
+
report={'status':'preparing','parent_revision':REV,'hardware':torch.cuda.get_device_name(0),'job_timeout_seconds':5400,'listed_usd_per_hour':1.0,'training_seed':20261113}
|
| 21 |
+
|
| 22 |
+
def publish():
|
| 23 |
+
report['elapsed_seconds']=round(time.monotonic()-started,2)
|
| 24 |
+
(OUT/'v3_report.json').write_text(json.dumps(report,indent=2))
|
| 25 |
+
api.upload_folder(repo_id=REPO,folder_path=OUT,revision=branch,path_in_repo='v3-results',ignore_patterns=['merged/**'],commit_message='Record measured V3 training and evaluation artifacts')
|
| 26 |
+
|
| 27 |
+
def sample(a,b,c,kind):
|
| 28 |
+
if kind==0:return f'Return only the integer: {a} + {b} - {c}.',str(a+b-c),'integer'
|
| 29 |
+
if kind==1:return f'A crate has {a} red and {b} blue beads. Remove {c} red beads. Return only the total beads remaining.',str(a+b-c),'integer'
|
| 30 |
+
if kind==2:return f'Return only valid JSON with keys "item" and "count": item=part_{a}, count={b}.',{'item':f'part_{a}','count':b},'json'
|
| 31 |
+
if kind==3:return f'Return only one CSV value row, no header: name=Item{a}; quantity={b}; code=K{c}.',f'Item{a},{b},K{c}','exact'
|
| 32 |
+
if kind==4:return f'Copy exactly, no other text: Key_{a}-B{b}_C{c}',f'Key_{a}-B{b}_C{c}','exact'
|
| 33 |
+
return f'Write only a Python function named solve(x) that returns x * {b} + {c}.',f'x * {b} + {c}','code'
|
| 34 |
+
|
| 35 |
+
def grade(case,text):
|
| 36 |
+
expected=case['expected']; kind=case['kind']; text=text.strip()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
try:
|
| 38 |
+
if kind=='json':return json.loads(text)==expected
|
| 39 |
+
if kind=='integer':return re.fullmatch(r'-?\d+',text) is not None and int(text)==int(expected)
|
| 40 |
+
if kind=='exact':return text==expected
|
| 41 |
+
# Parse and evaluate a tiny arithmetic-only AST; never execute model-generated code.
|
| 42 |
+
import ast
|
| 43 |
+
clean=re.sub(r'^```(?:python)?\s*|\s*```$','',text).strip();tree=ast.parse(clean)
|
| 44 |
+
if len(tree.body)!=1 or not isinstance(tree.body[0],ast.FunctionDef):return False
|
| 45 |
+
fn=tree.body[0]
|
| 46 |
+
if fn.name!='solve' or len(fn.args.args)!=1 or fn.args.args[0].arg!='x' or len(fn.body)!=1 or not isinstance(fn.body[0],ast.Return):return False
|
| 47 |
+
def calc(n,x):
|
| 48 |
+
if isinstance(n,ast.Constant) and isinstance(n.value,(int,float)):return n.value
|
| 49 |
+
if isinstance(n,ast.Name) and n.id=='x':return x
|
| 50 |
+
if isinstance(n,ast.BinOp) and isinstance(n.op,(ast.Add,ast.Sub,ast.Mult)):
|
| 51 |
+
a,b=calc(n.left,x),calc(n.right,x)
|
| 52 |
+
return a+b if isinstance(n.op,ast.Add) else a-b if isinstance(n.op,ast.Sub) else a*b
|
| 53 |
+
raise ValueError('Unsupported expression')
|
| 54 |
+
match=re.fullmatch(r'x \* (\d+) \+ (\d+)',expected);b,c=map(int,match.groups())
|
| 55 |
+
return all(calc(fn.body[0].value,x)==x*b+c for x in [-5,0,1,7,100])
|
| 56 |
+
except Exception:return False
|
| 57 |
+
|
| 58 |
+
tests=[]
|
| 59 |
+
for i in range(60):
|
| 60 |
+
# Evaluation numbers are outside the synthetic training ranges.
|
| 61 |
+
a,b,c=701+i*3,91+i,31+i%17;p,e,k=sample(a,b,c,i%6)
|
| 62 |
+
tests.append({'id':f'heldout_{i:02}','prompt':p,'expected':e,'kind':k})
|
| 63 |
+
train=[]
|
| 64 |
+
for i in range(900):
|
| 65 |
+
a,b,c=rng.randint(100,500),rng.randint(10,80),rng.randint(1,29)
|
| 66 |
+
p,e,k=sample(a,b,c,i%6)
|
| 67 |
+
answer=json.dumps(e) if k=='json' else 'def solve(x):\n return '+e if k=='code' else str(e)
|
| 68 |
+
train.append({'messages':[{'role':'user','content':p},{'role':'assistant','content':answer}]})
|
| 69 |
+
retained=[json.loads(s) for s in Path(hf_hub_download(REPO,'train.jsonl',revision=REV)).read_text().splitlines()]
|
| 70 |
+
rng.shuffle(retained);train+=retained[:900];rng.shuffle(train)
|
| 71 |
+
assert not {x['prompt'] for x in tests} & {m['content'] for row in train for m in row['messages'] if m['role']=='user'}
|
| 72 |
+
(OUT/'heldout_cases.json').write_text(json.dumps(tests,indent=2));(OUT/'training.jsonl').write_text('\n'.join(json.dumps(x) for x in train)+'\n')
|
| 73 |
+
(OUT/'train_v3.py').write_text(Path(__file__).read_text())
|
| 74 |
+
q=BitsAndBytesConfig(load_in_4bit=True,bnb_4bit_quant_type='nf4',bnb_4bit_use_double_quant=True,bnb_4bit_compute_dtype=torch.bfloat16)
|
| 75 |
+
model=AutoModelForCausalLM.from_pretrained(REPO,revision=REV,quantization_config=q,device_map={'':'cuda:0'},torch_dtype=torch.bfloat16,attn_implementation='sdpa')
|
| 76 |
+
|
| 77 |
@torch.inference_mode()
|
| 78 |
+
def evaluate(label):
|
| 79 |
+
model.eval(); rows=[]
|
| 80 |
+
for i,case in enumerate(tests):
|
| 81 |
+
inputs=tok.apply_chat_template([{'role':'system','content':settings['system']},{'role':'user','content':case['prompt']}],tokenize=True,add_generation_prompt=True,return_dict=True,return_tensors='pt').to('cuda')
|
| 82 |
+
out=model.generate(**inputs,max_new_tokens=96,do_sample=False,pad_token_id=128004,eos_token_id=128009)
|
| 83 |
+
text=tok.decode(out[0,inputs['input_ids'].shape[1]:],skip_special_tokens=True)
|
| 84 |
+
rows.append({**case,'output':text,'pass':grade(case,text),'tokens':int(out.shape[1]-inputs['input_ids'].shape[1])})
|
| 85 |
+
print('EVAL',label,i,rows[-1]['pass'],flush=True)
|
| 86 |
+
(OUT/(label+'_outputs.json')).write_text(json.dumps(rows,indent=2))
|
| 87 |
+
return rows
|
| 88 |
+
|
| 89 |
+
try:
|
| 90 |
+
baseline=evaluate('v2_quantized'); report['baseline_passed']=sum(r['pass'] for r in baseline);publish()
|
| 91 |
+
model=prepare_model_for_kbit_training(model,use_gradient_checkpointing=True)
|
| 92 |
+
model=get_peft_model(model,LoraConfig(r=16,lora_alpha=32,lora_dropout=0.05,target_modules=['q_proj','k_proj','v_proj','o_proj','gate_proj','up_proj','down_proj'],task_type='CAUSAL_LM'))
|
| 93 |
+
model.config.use_cache=False;encoded=[]
|
| 94 |
+
for row in train:
|
| 95 |
+
messages=row['messages']; prefix=tok.apply_chat_template(messages[:-1],tokenize=True,add_generation_prompt=True)
|
| 96 |
+
full=tok.apply_chat_template(messages,tokenize=True,add_generation_prompt=False)
|
| 97 |
+
if len(full)<=512 and full[:len(prefix)]==prefix:encoded.append({'input_ids':full,'labels':[-100]*len(prefix)+full[len(prefix):]})
|
| 98 |
+
def collate(rows):
|
| 99 |
+
length=max(len(r['input_ids']) for r in rows)
|
| 100 |
+
return {k:torch.tensor([r[k]+[pad]*(length-len(r[k])) for r in rows]) for k,pad in [('input_ids',128004),('labels',-100)]}|{'attention_mask':torch.tensor([[1]*len(r['input_ids'])+[0]*(length-len(r['input_ids'])) for r in rows])}
|
| 101 |
+
class Budget(TrainerCallback):
|
| 102 |
+
def on_step_end(self,args,state,control,**kwargs):
|
| 103 |
+
if time.monotonic()-started>2400:control.should_training_stop=True
|
| 104 |
+
return control
|
| 105 |
+
args=TrainingArguments(output_dir='/tmp/v3-checkpoints',max_steps=180,per_device_train_batch_size=1,gradient_accumulation_steps=4,learning_rate=1e-5,warmup_ratio=0.05,bf16=True,gradient_checkpointing=True,gradient_checkpointing_kwargs={'use_reentrant':False},optim='paged_adamw_8bit',save_strategy='no',logging_steps=10,report_to='none',remove_unused_columns=False,seed=20261113)
|
| 106 |
+
trainer=Trainer(model=model,args=args,train_dataset=encoded,data_collator=collate,callbacks=[Budget()])
|
| 107 |
+
report.update(status='training',training_examples=len(encoded),trainable_parameters=sum(p.numel() for p in model.parameters() if p.requires_grad));publish()
|
| 108 |
+
result=trainer.train();report['training']=result.metrics;report['optimizer_steps']=trainer.state.global_step
|
| 109 |
+
adapter=OUT/'adapter';model.save_pretrained(adapter,safe_serialization=True);tok.save_pretrained(adapter)
|
| 110 |
+
model.config.use_cache=True;model.gradient_checkpointing_disable();candidate=evaluate('v3_quantized')
|
| 111 |
+
report['candidate_passed']=sum(r['pass'] for r in candidate);report['status']='trained';publish()
|
| 112 |
+
del trainer,model;gc.collect();torch.cuda.empty_cache()
|
| 113 |
+
model=AutoModelForCausalLM.from_pretrained(REPO,revision=REV,torch_dtype=torch.bfloat16,device_map={'':'cuda:0'},attn_implementation='sdpa')
|
| 114 |
+
model=PeftModel.from_pretrained(model,adapter).merge_and_unload(safe_merge=True)
|
| 115 |
+
model.config.astheral_version='3';merged=OUT/'merged';model.save_pretrained(merged,safe_serialization=True,max_shard_size='4GB');tok.save_pretrained(merged)
|
| 116 |
+
del model;gc.collect();torch.cuda.empty_cache()
|
| 117 |
+
model=AutoModelForCausalLM.from_pretrained(merged,torch_dtype=torch.bfloat16,device_map={'':'cuda:0'},attn_implementation='sdpa')
|
| 118 |
+
reloaded=evaluate('v3_reloaded');report['reloaded_passed']=sum(r['pass'] for r in reloaded)
|
| 119 |
+
groups={kind:{'baseline':sum(r['pass'] for r in baseline if r['kind']==kind),'candidate':sum(r['pass'] for r in reloaded if r['kind']==kind),'total':sum(r['kind']==kind for r in tests)} for kind in {x['kind'] for x in tests}}
|
| 120 |
+
report['groups']=groups;report['quality_gate_passed']=report['reloaded_passed']>report['baseline_passed'] and all(v['candidate']>=v['baseline'] for v in groups.values())
|
| 121 |
+
report['status']='complete';report['peak_gpu_bytes']=torch.cuda.max_memory_allocated();publish()
|
| 122 |
+
# Save the actual trained weights to the isolated candidate branch even on a failed gate.
|
| 123 |
+
api.upload_folder(repo_id=REPO,folder_path=merged,revision=branch,commit_message='Save reloaded V3 candidate weights; see measured quality gate in v3-results')
|
| 124 |
+
if report['quality_gate_passed']:
|
| 125 |
+
api.create_tag(REPO,tag='v3',revision=api.model_info(REPO,revision=branch).sha,exist_ok=True)
|
| 126 |
+
print('V3_COMPLETE',json.dumps(report),flush=True)
|
| 127 |
+
except Exception:
|
| 128 |
+
report['status']='failed';report['error']=traceback.format_exc();publish();raise
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
trainer_log.json
CHANGED
|
@@ -1,230 +1,10 @@
|
|
| 1 |
[
|
| 2 |
{
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
| 8 |
-
|
| 9 |
-
{
|
| 10 |
-
"loss": 0.3021,
|
| 11 |
-
"grad_norm": 1.4435440301895142,
|
| 12 |
-
"learning_rate": 2.9992024290225105e-05,
|
| 13 |
-
"epoch": 0.09894867037724181,
|
| 14 |
-
"step": 20
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"loss": 0.2424,
|
| 18 |
-
"grad_norm": 2.773944854736328,
|
| 19 |
-
"learning_rate": 2.9850469719489576e-05,
|
| 20 |
-
"epoch": 0.14842300556586271,
|
| 21 |
-
"step": 30
|
| 22 |
-
},
|
| 23 |
-
{
|
| 24 |
-
"loss": 0.1925,
|
| 25 |
-
"grad_norm": 1.2467366456985474,
|
| 26 |
-
"learning_rate": 2.953360117250643e-05,
|
| 27 |
-
"epoch": 0.19789734075448362,
|
| 28 |
-
"step": 40
|
| 29 |
-
},
|
| 30 |
-
{
|
| 31 |
-
"loss": 0.1105,
|
| 32 |
-
"grad_norm": 0.9168190360069275,
|
| 33 |
-
"learning_rate": 2.9045159371688403e-05,
|
| 34 |
-
"epoch": 0.24737167594310452,
|
| 35 |
-
"step": 50
|
| 36 |
-
},
|
| 37 |
-
{
|
| 38 |
-
"loss": 0.1468,
|
| 39 |
-
"grad_norm": 1.8128526210784912,
|
| 40 |
-
"learning_rate": 2.8390910510016898e-05,
|
| 41 |
-
"epoch": 0.29684601113172543,
|
| 42 |
-
"step": 60
|
| 43 |
-
},
|
| 44 |
-
{
|
| 45 |
-
"loss": 0.1311,
|
| 46 |
-
"grad_norm": 1.2351698875427246,
|
| 47 |
-
"learning_rate": 2.7578578179510324e-05,
|
| 48 |
-
"epoch": 0.3463203463203463,
|
| 49 |
-
"step": 70
|
| 50 |
-
},
|
| 51 |
-
{
|
| 52 |
-
"loss": 0.1509,
|
| 53 |
-
"grad_norm": 5.074930667877197,
|
| 54 |
-
"learning_rate": 2.6617752192045723e-05,
|
| 55 |
-
"epoch": 0.39579468150896724,
|
| 56 |
-
"step": 80
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"loss": 0.1261,
|
| 60 |
-
"grad_norm": 1.543472409248352,
|
| 61 |
-
"learning_rate": 2.551977536892951e-05,
|
| 62 |
-
"epoch": 0.4452690166975881,
|
| 63 |
-
"step": 90
|
| 64 |
-
},
|
| 65 |
-
{
|
| 66 |
-
"loss": 0.1672,
|
| 67 |
-
"grad_norm": 1.9173252582550049,
|
| 68 |
-
"learning_rate": 2.4297609635698322e-05,
|
| 69 |
-
"epoch": 0.49474335188620905,
|
| 70 |
-
"step": 100
|
| 71 |
-
},
|
| 72 |
-
{
|
| 73 |
-
"loss": 0.1723,
|
| 74 |
-
"grad_norm": 0.7875851988792419,
|
| 75 |
-
"learning_rate": 2.29656830029386e-05,
|
| 76 |
-
"epoch": 0.54421768707483,
|
| 77 |
-
"step": 110
|
| 78 |
-
},
|
| 79 |
-
{
|
| 80 |
-
"loss": 0.1244,
|
| 81 |
-
"grad_norm": 1.0968589782714844,
|
| 82 |
-
"learning_rate": 2.153971923955934e-05,
|
| 83 |
-
"epoch": 0.5936920222634509,
|
| 84 |
-
"step": 120
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"loss": 0.1643,
|
| 88 |
-
"grad_norm": 1.792441725730896,
|
| 89 |
-
"learning_rate": 2.0036552249273017e-05,
|
| 90 |
-
"epoch": 0.6431663574520717,
|
| 91 |
-
"step": 130
|
| 92 |
-
},
|
| 93 |
-
{
|
| 94 |
-
"loss": 0.1358,
|
| 95 |
-
"grad_norm": 1.13428795337677,
|
| 96 |
-
"learning_rate": 1.8473927341622628e-05,
|
| 97 |
-
"epoch": 0.6926406926406926,
|
| 98 |
-
"step": 140
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"loss": 0.1403,
|
| 102 |
-
"grad_norm": 0.8378678560256958,
|
| 103 |
-
"learning_rate": 1.6870291743606273e-05,
|
| 104 |
-
"epoch": 0.7421150278293135,
|
| 105 |
-
"step": 150
|
| 106 |
-
},
|
| 107 |
-
{
|
| 108 |
-
"loss": 0.1757,
|
| 109 |
-
"grad_norm": 0.8258570432662964,
|
| 110 |
-
"learning_rate": 1.5244576824968541e-05,
|
| 111 |
-
"epoch": 0.7915893630179345,
|
| 112 |
-
"step": 160
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"loss": 0.2255,
|
| 116 |
-
"grad_norm": 1.775040864944458,
|
| 117 |
-
"learning_rate": 1.3615974608050472e-05,
|
| 118 |
-
"epoch": 0.8410636982065554,
|
| 119 |
-
"step": 170
|
| 120 |
-
},
|
| 121 |
-
{
|
| 122 |
-
"loss": 0.1666,
|
| 123 |
-
"grad_norm": 1.0309557914733887,
|
| 124 |
-
"learning_rate": 1.2003711200562243e-05,
|
| 125 |
-
"epoch": 0.8905380333951762,
|
| 126 |
-
"step": 180
|
| 127 |
-
},
|
| 128 |
-
{
|
| 129 |
-
"loss": 0.113,
|
| 130 |
-
"grad_norm": 0.45190468430519104,
|
| 131 |
-
"learning_rate": 1.042681982596845e-05,
|
| 132 |
-
"epoch": 0.9400123685837971,
|
| 133 |
-
"step": 190
|
| 134 |
-
},
|
| 135 |
-
{
|
| 136 |
-
"loss": 0.1376,
|
| 137 |
-
"grad_norm": 1.603728175163269,
|
| 138 |
-
"learning_rate": 8.903916130926232e-06,
|
| 139 |
-
"epoch": 0.9894867037724181,
|
| 140 |
-
"step": 200
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"loss": 0.0922,
|
| 144 |
-
"grad_norm": 0.7828471064567566,
|
| 145 |
-
"learning_rate": 7.452978422335191e-06,
|
| 146 |
-
"epoch": 1.0346320346320346,
|
| 147 |
-
"step": 210
|
| 148 |
-
},
|
| 149 |
-
{
|
| 150 |
-
"loss": 0.0946,
|
| 151 |
-
"grad_norm": 0.4640823304653168,
|
| 152 |
-
"learning_rate": 6.091135428362536e-06,
|
| 153 |
-
"epoch": 1.0841063698206554,
|
| 154 |
-
"step": 220
|
| 155 |
-
},
|
| 156 |
-
{
|
| 157 |
-
"loss": 0.0761,
|
| 158 |
-
"grad_norm": 0.7620387077331543,
|
| 159 |
-
"learning_rate": 4.834464088984452e-06,
|
| 160 |
-
"epoch": 1.1335807050092765,
|
| 161 |
-
"step": 230
|
| 162 |
-
},
|
| 163 |
-
{
|
| 164 |
-
"loss": 0.0995,
|
| 165 |
-
"grad_norm": 0.5594423413276672,
|
| 166 |
-
"learning_rate": 3.697799763183127e-06,
|
| 167 |
-
"epoch": 1.1830550401978974,
|
| 168 |
-
"step": 240
|
| 169 |
-
},
|
| 170 |
-
{
|
| 171 |
-
"loss": 0.0716,
|
| 172 |
-
"grad_norm": 0.008049696683883667,
|
| 173 |
-
"learning_rate": 2.6945610933570297e-06,
|
| 174 |
-
"epoch": 1.2325293753865183,
|
| 175 |
-
"step": 250
|
| 176 |
-
},
|
| 177 |
-
{
|
| 178 |
-
"loss": 0.0468,
|
| 179 |
-
"grad_norm": 1.0049060583114624,
|
| 180 |
-
"learning_rate": 1.8365915944693102e-06,
|
| 181 |
-
"epoch": 1.2820037105751392,
|
| 182 |
-
"step": 260
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"loss": 0.0501,
|
| 186 |
-
"grad_norm": 0.658693790435791,
|
| 187 |
-
"learning_rate": 1.1340198380190725e-06,
|
| 188 |
-
"epoch": 1.33147804576376,
|
| 189 |
-
"step": 270
|
| 190 |
-
},
|
| 191 |
-
{
|
| 192 |
-
"loss": 0.138,
|
| 193 |
-
"grad_norm": 1.3238775730133057,
|
| 194 |
-
"learning_rate": 5.951398814030951e-07,
|
| 195 |
-
"epoch": 1.380952380952381,
|
| 196 |
-
"step": 280
|
| 197 |
-
},
|
| 198 |
-
{
|
| 199 |
-
"loss": 0.0941,
|
| 200 |
-
"grad_norm": 0.4109145700931549,
|
| 201 |
-
"learning_rate": 2.2631335423303312e-07,
|
| 202 |
-
"epoch": 1.4304267161410018,
|
| 203 |
-
"step": 290
|
| 204 |
-
},
|
| 205 |
-
{
|
| 206 |
-
"loss": 0.0745,
|
| 207 |
-
"grad_norm": 0.4245922863483429,
|
| 208 |
-
"learning_rate": 3.189435750671321e-08,
|
| 209 |
-
"epoch": 1.4799010513296227,
|
| 210 |
-
"step": 300
|
| 211 |
-
},
|
| 212 |
-
{
|
| 213 |
-
"train_runtime": 1007.3112,
|
| 214 |
-
"train_samples_per_second": 2.408,
|
| 215 |
-
"train_steps_per_second": 0.303,
|
| 216 |
-
"total_flos": 2.0134086292832256e+16,
|
| 217 |
-
"train_loss": 0.15099767006811549,
|
| 218 |
-
"epoch": 1.5046382189239331,
|
| 219 |
-
"step": 305
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"eval_loss": 0.0020501785911619663,
|
| 223 |
-
"eval_model_preparation_time": 0.0184,
|
| 224 |
-
"eval_runtime": 5.753,
|
| 225 |
-
"eval_samples_per_second": 11.125,
|
| 226 |
-
"eval_steps_per_second": 11.125,
|
| 227 |
-
"epoch": 1.5046382189239331,
|
| 228 |
-
"step": 305
|
| 229 |
}
|
| 230 |
]
|
|
|
|
| 1 |
[
|
| 2 |
{
|
| 3 |
+
"train_runtime": 390.2404,
|
| 4 |
+
"train_samples_per_second": 1.845,
|
| 5 |
+
"train_steps_per_second": 0.461,
|
| 6 |
+
"total_flos": 3960249019342848.0,
|
| 7 |
+
"train_loss": 0.2867420093880759,
|
| 8 |
+
"epoch": 0.4203152364273205
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
}
|
| 10 |
]
|
v3-results/adapter/README.md
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: SEE7/Astheral
|
| 3 |
+
library_name: peft
|
| 4 |
+
pipeline_tag: text-generation
|
| 5 |
+
tags:
|
| 6 |
+
- base_model:adapter:SEE7/Astheral
|
| 7 |
+
- lora
|
| 8 |
+
- transformers
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Model ID
|
| 12 |
+
|
| 13 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
## Model Details
|
| 18 |
+
|
| 19 |
+
### Model Description
|
| 20 |
+
|
| 21 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
- **Developed by:** [More Information Needed]
|
| 26 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 27 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 28 |
+
- **Model type:** [More Information Needed]
|
| 29 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 30 |
+
- **License:** [More Information Needed]
|
| 31 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 32 |
+
|
| 33 |
+
### Model Sources [optional]
|
| 34 |
+
|
| 35 |
+
<!-- Provide the basic links for the model. -->
|
| 36 |
+
|
| 37 |
+
- **Repository:** [More Information Needed]
|
| 38 |
+
- **Paper [optional]:** [More Information Needed]
|
| 39 |
+
- **Demo [optional]:** [More Information Needed]
|
| 40 |
+
|
| 41 |
+
## Uses
|
| 42 |
+
|
| 43 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 44 |
+
|
| 45 |
+
### Direct Use
|
| 46 |
+
|
| 47 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 48 |
+
|
| 49 |
+
[More Information Needed]
|
| 50 |
+
|
| 51 |
+
### Downstream Use [optional]
|
| 52 |
+
|
| 53 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 54 |
+
|
| 55 |
+
[More Information Needed]
|
| 56 |
+
|
| 57 |
+
### Out-of-Scope Use
|
| 58 |
+
|
| 59 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 60 |
+
|
| 61 |
+
[More Information Needed]
|
| 62 |
+
|
| 63 |
+
## Bias, Risks, and Limitations
|
| 64 |
+
|
| 65 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 66 |
+
|
| 67 |
+
[More Information Needed]
|
| 68 |
+
|
| 69 |
+
### Recommendations
|
| 70 |
+
|
| 71 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 72 |
+
|
| 73 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 74 |
+
|
| 75 |
+
## How to Get Started with the Model
|
| 76 |
+
|
| 77 |
+
Use the code below to get started with the model.
|
| 78 |
+
|
| 79 |
+
[More Information Needed]
|
| 80 |
+
|
| 81 |
+
## Training Details
|
| 82 |
+
|
| 83 |
+
### Training Data
|
| 84 |
+
|
| 85 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 86 |
+
|
| 87 |
+
[More Information Needed]
|
| 88 |
+
|
| 89 |
+
### Training Procedure
|
| 90 |
+
|
| 91 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 92 |
+
|
| 93 |
+
#### Preprocessing [optional]
|
| 94 |
+
|
| 95 |
+
[More Information Needed]
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
#### Training Hyperparameters
|
| 99 |
+
|
| 100 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 101 |
+
|
| 102 |
+
#### Speeds, Sizes, Times [optional]
|
| 103 |
+
|
| 104 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 105 |
+
|
| 106 |
+
[More Information Needed]
|
| 107 |
+
|
| 108 |
+
## Evaluation
|
| 109 |
+
|
| 110 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 111 |
+
|
| 112 |
+
### Testing Data, Factors & Metrics
|
| 113 |
+
|
| 114 |
+
#### Testing Data
|
| 115 |
+
|
| 116 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 117 |
+
|
| 118 |
+
[More Information Needed]
|
| 119 |
+
|
| 120 |
+
#### Factors
|
| 121 |
+
|
| 122 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 123 |
+
|
| 124 |
+
[More Information Needed]
|
| 125 |
+
|
| 126 |
+
#### Metrics
|
| 127 |
+
|
| 128 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 129 |
+
|
| 130 |
+
[More Information Needed]
|
| 131 |
+
|
| 132 |
+
### Results
|
| 133 |
+
|
| 134 |
+
[More Information Needed]
|
| 135 |
+
|
| 136 |
+
#### Summary
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
|
| 140 |
+
## Model Examination [optional]
|
| 141 |
+
|
| 142 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 143 |
+
|
| 144 |
+
[More Information Needed]
|
| 145 |
+
|
| 146 |
+
## Environmental Impact
|
| 147 |
+
|
| 148 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 149 |
+
|
| 150 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 151 |
+
|
| 152 |
+
- **Hardware Type:** [More Information Needed]
|
| 153 |
+
- **Hours used:** [More Information Needed]
|
| 154 |
+
- **Cloud Provider:** [More Information Needed]
|
| 155 |
+
- **Compute Region:** [More Information Needed]
|
| 156 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 157 |
+
|
| 158 |
+
## Technical Specifications [optional]
|
| 159 |
+
|
| 160 |
+
### Model Architecture and Objective
|
| 161 |
+
|
| 162 |
+
[More Information Needed]
|
| 163 |
+
|
| 164 |
+
### Compute Infrastructure
|
| 165 |
+
|
| 166 |
+
[More Information Needed]
|
| 167 |
+
|
| 168 |
+
#### Hardware
|
| 169 |
+
|
| 170 |
+
[More Information Needed]
|
| 171 |
+
|
| 172 |
+
#### Software
|
| 173 |
+
|
| 174 |
+
[More Information Needed]
|
| 175 |
+
|
| 176 |
+
## Citation [optional]
|
| 177 |
+
|
| 178 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 179 |
+
|
| 180 |
+
**BibTeX:**
|
| 181 |
+
|
| 182 |
+
[More Information Needed]
|
| 183 |
+
|
| 184 |
+
**APA:**
|
| 185 |
+
|
| 186 |
+
[More Information Needed]
|
| 187 |
+
|
| 188 |
+
## Glossary [optional]
|
| 189 |
+
|
| 190 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 191 |
+
|
| 192 |
+
[More Information Needed]
|
| 193 |
+
|
| 194 |
+
## More Information [optional]
|
| 195 |
+
|
| 196 |
+
[More Information Needed]
|
| 197 |
+
|
| 198 |
+
## Model Card Authors [optional]
|
| 199 |
+
|
| 200 |
+
[More Information Needed]
|
| 201 |
+
|
| 202 |
+
## Model Card Contact
|
| 203 |
+
|
| 204 |
+
[More Information Needed]
|
| 205 |
+
### Framework versions
|
| 206 |
+
|
| 207 |
+
- PEFT 0.17.1
|
v3-results/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "SEE7/Astheral",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"corda_config": null,
|
| 7 |
+
"eva_config": null,
|
| 8 |
+
"exclude_modules": null,
|
| 9 |
+
"fan_in_fan_out": false,
|
| 10 |
+
"inference_mode": true,
|
| 11 |
+
"init_lora_weights": true,
|
| 12 |
+
"layer_replication": null,
|
| 13 |
+
"layers_pattern": null,
|
| 14 |
+
"layers_to_transform": null,
|
| 15 |
+
"loftq_config": {},
|
| 16 |
+
"lora_alpha": 32,
|
| 17 |
+
"lora_bias": false,
|
| 18 |
+
"lora_dropout": 0.05,
|
| 19 |
+
"megatron_config": null,
|
| 20 |
+
"megatron_core": "megatron.core",
|
| 21 |
+
"modules_to_save": null,
|
| 22 |
+
"peft_type": "LORA",
|
| 23 |
+
"qalora_group_size": 16,
|
| 24 |
+
"r": 16,
|
| 25 |
+
"rank_pattern": {},
|
| 26 |
+
"revision": "d9aa899722be4788f57107c2951c72d0ec6614ee",
|
| 27 |
+
"target_modules": [
|
| 28 |
+
"gate_proj",
|
| 29 |
+
"o_proj",
|
| 30 |
+
"up_proj",
|
| 31 |
+
"down_proj",
|
| 32 |
+
"v_proj",
|
| 33 |
+
"q_proj",
|
| 34 |
+
"k_proj"
|
| 35 |
+
],
|
| 36 |
+
"target_parameters": null,
|
| 37 |
+
"task_type": "CAUSAL_LM",
|
| 38 |
+
"trainable_token_indices": null,
|
| 39 |
+
"use_dora": false,
|
| 40 |
+
"use_qalora": false,
|
| 41 |
+
"use_rslora": false
|
| 42 |
+
}
|
v3-results/adapter/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d9a26fcd5590725a2cfdfec6246c93ea91e237e3bf34e214b366b2387feb00ae
|
| 3 |
+
size 167832240
|
v3-results/adapter/additional_chat_templates/tool_use.jinja
ADDED
|
@@ -0,0 +1,152 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- macro json_to_python_type(json_spec) %}
|
| 2 |
+
{%- set basic_type_map = {
|
| 3 |
+
"string": "str",
|
| 4 |
+
"number": "float",
|
| 5 |
+
"integer": "int",
|
| 6 |
+
"boolean": "bool"
|
| 7 |
+
} %}
|
| 8 |
+
|
| 9 |
+
{%- if basic_type_map[json_spec.type] is defined %}
|
| 10 |
+
{{- basic_type_map[json_spec.type] }}
|
| 11 |
+
{%- elif json_spec.type == "array" %}
|
| 12 |
+
{{- "list[" + json_to_python_type(json_spec|items) + "]"}}
|
| 13 |
+
{%- elif json_spec.type == "object" %}
|
| 14 |
+
{%- if json_spec.additionalProperties is defined %}
|
| 15 |
+
{{- "dict[str, " + json_to_python_type(json_spec.additionalProperties) + ']'}}
|
| 16 |
+
{%- else %}
|
| 17 |
+
{{- "dict" }}
|
| 18 |
+
{%- endif %}
|
| 19 |
+
{%- elif json_spec.type is iterable %}
|
| 20 |
+
{{- "Union[" }}
|
| 21 |
+
{%- for t in json_spec.type %}
|
| 22 |
+
{{- json_to_python_type({"type": t}) }}
|
| 23 |
+
{%- if not loop.last %}
|
| 24 |
+
{{- "," }}
|
| 25 |
+
{%- endif %}
|
| 26 |
+
{%- endfor %}
|
| 27 |
+
{{- "]" }}
|
| 28 |
+
{%- else %}
|
| 29 |
+
{{- "Any" }}
|
| 30 |
+
{%- endif %}
|
| 31 |
+
{%- endmacro %}
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
{{- bos_token }}
|
| 35 |
+
{{- '<|im_start|>system
|
| 36 |
+
' }}
|
| 37 |
+
{{- "You are a function calling AI model. You are provided with function signatures within <tools></tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions. Here are the available tools: <tools> " }}
|
| 38 |
+
{%- for tool in tools %}
|
| 39 |
+
{%- if tool.function is defined %}
|
| 40 |
+
{%- set tool = tool.function %}
|
| 41 |
+
{%- endif %}
|
| 42 |
+
{{- '{"type": "function", "function": ' }}
|
| 43 |
+
{{- '{"name": "' + tool.name + '", ' }}
|
| 44 |
+
{{- '"description": "' + tool.name + '(' }}
|
| 45 |
+
{%- for param_name, param_fields in tool.parameters.properties|items %}
|
| 46 |
+
{{- param_name + ": " + json_to_python_type(param_fields) }}
|
| 47 |
+
{%- if not loop.last %}
|
| 48 |
+
{{- ", " }}
|
| 49 |
+
{%- endif %}
|
| 50 |
+
{%- endfor %}
|
| 51 |
+
{{- ")" }}
|
| 52 |
+
{%- if tool.return is defined %}
|
| 53 |
+
{{- " -> " + json_to_python_type(tool.return) }}
|
| 54 |
+
{%- endif %}
|
| 55 |
+
{{- " - " + tool.description + "
|
| 56 |
+
|
| 57 |
+
" }}
|
| 58 |
+
{%- for param_name, param_fields in tool.parameters.properties|items %}
|
| 59 |
+
{%- if loop.first %}
|
| 60 |
+
{{- " Args:
|
| 61 |
+
" }}
|
| 62 |
+
{%- endif %}
|
| 63 |
+
{{- " " + param_name + "(" + json_to_python_type(param_fields) + "): " + param_fields.description|trim }}
|
| 64 |
+
{%- endfor %}
|
| 65 |
+
{%- if tool.return is defined and tool.return.description is defined %}
|
| 66 |
+
{{- "
|
| 67 |
+
Returns:
|
| 68 |
+
" + tool.return.description }}
|
| 69 |
+
{%- endif %}
|
| 70 |
+
{{- '"' }}
|
| 71 |
+
{{- ', "parameters": ' }}
|
| 72 |
+
{%- if tool.parameters.properties | length == 0 %}
|
| 73 |
+
{{- "{}" }}
|
| 74 |
+
{%- else %}
|
| 75 |
+
{{- tool.parameters|tojson }}
|
| 76 |
+
{%- endif %}
|
| 77 |
+
{{- "}" }}
|
| 78 |
+
{%- if not loop.last %}
|
| 79 |
+
{{- "
|
| 80 |
+
" }}
|
| 81 |
+
{%- endif %}
|
| 82 |
+
{%- endfor %}
|
| 83 |
+
{{- " </tools>" }}
|
| 84 |
+
{{- 'Use the following pydantic model json schema for each tool call you will make: {"properties": {"name": {"title": "Name", "type": "string"}, "arguments": {"title": "Arguments", "type": "object"}}, "required": ["name", "arguments"], "title": "FunctionCall", "type": "object"}}
|
| 85 |
+
' }}
|
| 86 |
+
{{- "For each function call return a json object with function name and arguments within <tool_call></tool_call> XML tags as follows:
|
| 87 |
+
" }}
|
| 88 |
+
{{- "<tool_call>
|
| 89 |
+
" }}
|
| 90 |
+
{{- '{"name": <function-name>, "arguments": <args-dict>}
|
| 91 |
+
' }}
|
| 92 |
+
{{- '</tool_call><|im_end|>
|
| 93 |
+
' }}
|
| 94 |
+
{%- for message in messages %}
|
| 95 |
+
{%- if message.role == "user" or message.role == "system" or (message.role == "assistant" and message.tool_calls is not defined) %}
|
| 96 |
+
{{- '<|im_start|>' + message.role + '
|
| 97 |
+
' + message.content + '<|im_end|>' + '
|
| 98 |
+
' }}
|
| 99 |
+
{%- elif message.role == "assistant" %}
|
| 100 |
+
{{- '<|im_start|>' + message.role }}
|
| 101 |
+
{%- for tool_call in message.tool_calls %}
|
| 102 |
+
{{- '
|
| 103 |
+
<tool_call>
|
| 104 |
+
' }} {%- if tool_call.function is defined %}
|
| 105 |
+
{%- set tool_call = tool_call.function %}
|
| 106 |
+
{%- endif %}
|
| 107 |
+
{{- '{' }}
|
| 108 |
+
{{- '"name": "' }}
|
| 109 |
+
{{- tool_call.name }}
|
| 110 |
+
{{- '"' }}
|
| 111 |
+
{{- ', '}}
|
| 112 |
+
{%- if tool_call.arguments is defined %}
|
| 113 |
+
{{- '"arguments": ' }}
|
| 114 |
+
{%- if tool_call.arguments is string %}
|
| 115 |
+
{{- tool_call.arguments }}
|
| 116 |
+
{%- else %}
|
| 117 |
+
{{- tool_call.arguments|tojson }}
|
| 118 |
+
{%- endif %}
|
| 119 |
+
{%- endif %}
|
| 120 |
+
{{- '}' }}
|
| 121 |
+
{{- '
|
| 122 |
+
</tool_call>' }}
|
| 123 |
+
{%- endfor %}
|
| 124 |
+
{{- '<|im_end|>
|
| 125 |
+
' }}
|
| 126 |
+
{%- elif message.role == "tool" %}
|
| 127 |
+
{%- if loop.previtem and loop.previtem.role != "tool" %}
|
| 128 |
+
{{- '<|im_start|>tool
|
| 129 |
+
' }}
|
| 130 |
+
{%- endif %}
|
| 131 |
+
{{- '<tool_response>
|
| 132 |
+
' }}
|
| 133 |
+
{{- message.content }}
|
| 134 |
+
{%- if not loop.last %}
|
| 135 |
+
{{- '
|
| 136 |
+
</tool_response>
|
| 137 |
+
' }}
|
| 138 |
+
{%- else %}
|
| 139 |
+
{{- '
|
| 140 |
+
</tool_response>' }}
|
| 141 |
+
{%- endif %}
|
| 142 |
+
{%- if not loop.last and loop.nextitem.role != "tool" %}
|
| 143 |
+
{{- '<|im_end|>' }}
|
| 144 |
+
{%- elif loop.last %}
|
| 145 |
+
{{- '<|im_end|>' }}
|
| 146 |
+
{%- endif %}
|
| 147 |
+
{%- endif %}
|
| 148 |
+
{%- endfor %}
|
| 149 |
+
{%- if add_generation_prompt %}
|
| 150 |
+
{{- '<|im_start|>assistant
|
| 151 |
+
' }}
|
| 152 |
+
{%- endif %}
|
v3-results/adapter/chat_template.jinja
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>
|
| 2 |
+
|
| 3 |
+
'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{% if add_generation_prompt %}{{ '<|start_header_id|>assistant<|end_header_id|>
|
| 4 |
+
|
| 5 |
+
' }}{% endif %}
|
v3-results/adapter/special_tokens_map.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<|begin_of_text|>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|eot_id|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": "<|finetune_right_pad_id|>"
|
| 17 |
+
}
|
v3-results/adapter/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
|
| 3 |
+
size 17209920
|
v3-results/adapter/tokenizer_config.json
ADDED
|
@@ -0,0 +1,2069 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"128000": {
|
| 5 |
+
"content": "<|begin_of_text|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"128001": {
|
| 13 |
+
"content": "<|end_of_text|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"128002": {
|
| 21 |
+
"content": "<|reserved_special_token_0|>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": false,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"128003": {
|
| 29 |
+
"content": "<|reserved_special_token_1|>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"128004": {
|
| 37 |
+
"content": "<|finetune_right_pad_id|>",
|
| 38 |
+
"lstrip": false,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
},
|
| 44 |
+
"128005": {
|
| 45 |
+
"content": "<|reserved_special_token_2|>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": false,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false,
|
| 50 |
+
"special": true
|
| 51 |
+
},
|
| 52 |
+
"128006": {
|
| 53 |
+
"content": "<|start_header_id|>",
|
| 54 |
+
"lstrip": false,
|
| 55 |
+
"normalized": false,
|
| 56 |
+
"rstrip": false,
|
| 57 |
+
"single_word": false,
|
| 58 |
+
"special": true
|
| 59 |
+
},
|
| 60 |
+
"128007": {
|
| 61 |
+
"content": "<|end_header_id|>",
|
| 62 |
+
"lstrip": false,
|
| 63 |
+
"normalized": false,
|
| 64 |
+
"rstrip": false,
|
| 65 |
+
"single_word": false,
|
| 66 |
+
"special": true
|
| 67 |
+
},
|
| 68 |
+
"128008": {
|
| 69 |
+
"content": "<|eom_id|>",
|
| 70 |
+
"lstrip": false,
|
| 71 |
+
"normalized": false,
|
| 72 |
+
"rstrip": false,
|
| 73 |
+
"single_word": false,
|
| 74 |
+
"special": true
|
| 75 |
+
},
|
| 76 |
+
"128009": {
|
| 77 |
+
"content": "<|eot_id|>",
|
| 78 |
+
"lstrip": false,
|
| 79 |
+
"normalized": false,
|
| 80 |
+
"rstrip": false,
|
| 81 |
+
"single_word": false,
|
| 82 |
+
"special": true
|
| 83 |
+
},
|
| 84 |
+
"128010": {
|
| 85 |
+
"content": "<|python_tag|>",
|
| 86 |
+
"lstrip": false,
|
| 87 |
+
"normalized": false,
|
| 88 |
+
"rstrip": false,
|
| 89 |
+
"single_word": false,
|
| 90 |
+
"special": true
|
| 91 |
+
},
|
| 92 |
+
"128011": {
|
| 93 |
+
"content": "<|reserved_special_token_3|>",
|
| 94 |
+
"lstrip": false,
|
| 95 |
+
"normalized": false,
|
| 96 |
+
"rstrip": false,
|
| 97 |
+
"single_word": false,
|
| 98 |
+
"special": true
|
| 99 |
+
},
|
| 100 |
+
"128012": {
|
| 101 |
+
"content": "<|reserved_special_token_4|>",
|
| 102 |
+
"lstrip": false,
|
| 103 |
+
"normalized": false,
|
| 104 |
+
"rstrip": false,
|
| 105 |
+
"single_word": false,
|
| 106 |
+
"special": true
|
| 107 |
+
},
|
| 108 |
+
"128013": {
|
| 109 |
+
"content": "<|reserved_special_token_5|>",
|
| 110 |
+
"lstrip": false,
|
| 111 |
+
"normalized": false,
|
| 112 |
+
"rstrip": false,
|
| 113 |
+
"single_word": false,
|
| 114 |
+
"special": true
|
| 115 |
+
},
|
| 116 |
+
"128014": {
|
| 117 |
+
"content": "<|reserved_special_token_6|>",
|
| 118 |
+
"lstrip": false,
|
| 119 |
+
"normalized": false,
|
| 120 |
+
"rstrip": false,
|
| 121 |
+
"single_word": false,
|
| 122 |
+
"special": true
|
| 123 |
+
},
|
| 124 |
+
"128015": {
|
| 125 |
+
"content": "<|reserved_special_token_7|>",
|
| 126 |
+
"lstrip": false,
|
| 127 |
+
"normalized": false,
|
| 128 |
+
"rstrip": false,
|
| 129 |
+
"single_word": false,
|
| 130 |
+
"special": true
|
| 131 |
+
},
|
| 132 |
+
"128016": {
|
| 133 |
+
"content": "<|reserved_special_token_8|>",
|
| 134 |
+
"lstrip": false,
|
| 135 |
+
"normalized": false,
|
| 136 |
+
"rstrip": false,
|
| 137 |
+
"single_word": false,
|
| 138 |
+
"special": true
|
| 139 |
+
},
|
| 140 |
+
"128017": {
|
| 141 |
+
"content": "<|reserved_special_token_9|>",
|
| 142 |
+
"lstrip": false,
|
| 143 |
+
"normalized": false,
|
| 144 |
+
"rstrip": false,
|
| 145 |
+
"single_word": false,
|
| 146 |
+
"special": true
|
| 147 |
+
},
|
| 148 |
+
"128018": {
|
| 149 |
+
"content": "<|reserved_special_token_10|>",
|
| 150 |
+
"lstrip": false,
|
| 151 |
+
"normalized": false,
|
| 152 |
+
"rstrip": false,
|
| 153 |
+
"single_word": false,
|
| 154 |
+
"special": true
|
| 155 |
+
},
|
| 156 |
+
"128019": {
|
| 157 |
+
"content": "<|reserved_special_token_11|>",
|
| 158 |
+
"lstrip": false,
|
| 159 |
+
"normalized": false,
|
| 160 |
+
"rstrip": false,
|
| 161 |
+
"single_word": false,
|
| 162 |
+
"special": true
|
| 163 |
+
},
|
| 164 |
+
"128020": {
|
| 165 |
+
"content": "<|reserved_special_token_12|>",
|
| 166 |
+
"lstrip": false,
|
| 167 |
+
"normalized": false,
|
| 168 |
+
"rstrip": false,
|
| 169 |
+
"single_word": false,
|
| 170 |
+
"special": true
|
| 171 |
+
},
|
| 172 |
+
"128021": {
|
| 173 |
+
"content": "<|reserved_special_token_13|>",
|
| 174 |
+
"lstrip": false,
|
| 175 |
+
"normalized": false,
|
| 176 |
+
"rstrip": false,
|
| 177 |
+
"single_word": false,
|
| 178 |
+
"special": true
|
| 179 |
+
},
|
| 180 |
+
"128022": {
|
| 181 |
+
"content": "<|reserved_special_token_14|>",
|
| 182 |
+
"lstrip": false,
|
| 183 |
+
"normalized": false,
|
| 184 |
+
"rstrip": false,
|
| 185 |
+
"single_word": false,
|
| 186 |
+
"special": true
|
| 187 |
+
},
|
| 188 |
+
"128023": {
|
| 189 |
+
"content": "<|reserved_special_token_15|>",
|
| 190 |
+
"lstrip": false,
|
| 191 |
+
"normalized": false,
|
| 192 |
+
"rstrip": false,
|
| 193 |
+
"single_word": false,
|
| 194 |
+
"special": true
|
| 195 |
+
},
|
| 196 |
+
"128024": {
|
| 197 |
+
"content": "<|reserved_special_token_16|>",
|
| 198 |
+
"lstrip": false,
|
| 199 |
+
"normalized": false,
|
| 200 |
+
"rstrip": false,
|
| 201 |
+
"single_word": false,
|
| 202 |
+
"special": true
|
| 203 |
+
},
|
| 204 |
+
"128025": {
|
| 205 |
+
"content": "<|reserved_special_token_17|>",
|
| 206 |
+
"lstrip": false,
|
| 207 |
+
"normalized": false,
|
| 208 |
+
"rstrip": false,
|
| 209 |
+
"single_word": false,
|
| 210 |
+
"special": true
|
| 211 |
+
},
|
| 212 |
+
"128026": {
|
| 213 |
+
"content": "<|reserved_special_token_18|>",
|
| 214 |
+
"lstrip": false,
|
| 215 |
+
"normalized": false,
|
| 216 |
+
"rstrip": false,
|
| 217 |
+
"single_word": false,
|
| 218 |
+
"special": true
|
| 219 |
+
},
|
| 220 |
+
"128027": {
|
| 221 |
+
"content": "<|reserved_special_token_19|>",
|
| 222 |
+
"lstrip": false,
|
| 223 |
+
"normalized": false,
|
| 224 |
+
"rstrip": false,
|
| 225 |
+
"single_word": false,
|
| 226 |
+
"special": true
|
| 227 |
+
},
|
| 228 |
+
"128028": {
|
| 229 |
+
"content": "<|reserved_special_token_20|>",
|
| 230 |
+
"lstrip": false,
|
| 231 |
+
"normalized": false,
|
| 232 |
+
"rstrip": false,
|
| 233 |
+
"single_word": false,
|
| 234 |
+
"special": true
|
| 235 |
+
},
|
| 236 |
+
"128029": {
|
| 237 |
+
"content": "<|reserved_special_token_21|>",
|
| 238 |
+
"lstrip": false,
|
| 239 |
+
"normalized": false,
|
| 240 |
+
"rstrip": false,
|
| 241 |
+
"single_word": false,
|
| 242 |
+
"special": true
|
| 243 |
+
},
|
| 244 |
+
"128030": {
|
| 245 |
+
"content": "<|reserved_special_token_22|>",
|
| 246 |
+
"lstrip": false,
|
| 247 |
+
"normalized": false,
|
| 248 |
+
"rstrip": false,
|
| 249 |
+
"single_word": false,
|
| 250 |
+
"special": true
|
| 251 |
+
},
|
| 252 |
+
"128031": {
|
| 253 |
+
"content": "<|reserved_special_token_23|>",
|
| 254 |
+
"lstrip": false,
|
| 255 |
+
"normalized": false,
|
| 256 |
+
"rstrip": false,
|
| 257 |
+
"single_word": false,
|
| 258 |
+
"special": true
|
| 259 |
+
},
|
| 260 |
+
"128032": {
|
| 261 |
+
"content": "<|reserved_special_token_24|>",
|
| 262 |
+
"lstrip": false,
|
| 263 |
+
"normalized": false,
|
| 264 |
+
"rstrip": false,
|
| 265 |
+
"single_word": false,
|
| 266 |
+
"special": true
|
| 267 |
+
},
|
| 268 |
+
"128033": {
|
| 269 |
+
"content": "<|reserved_special_token_25|>",
|
| 270 |
+
"lstrip": false,
|
| 271 |
+
"normalized": false,
|
| 272 |
+
"rstrip": false,
|
| 273 |
+
"single_word": false,
|
| 274 |
+
"special": true
|
| 275 |
+
},
|
| 276 |
+
"128034": {
|
| 277 |
+
"content": "<|reserved_special_token_26|>",
|
| 278 |
+
"lstrip": false,
|
| 279 |
+
"normalized": false,
|
| 280 |
+
"rstrip": false,
|
| 281 |
+
"single_word": false,
|
| 282 |
+
"special": true
|
| 283 |
+
},
|
| 284 |
+
"128035": {
|
| 285 |
+
"content": "<|reserved_special_token_27|>",
|
| 286 |
+
"lstrip": false,
|
| 287 |
+
"normalized": false,
|
| 288 |
+
"rstrip": false,
|
| 289 |
+
"single_word": false,
|
| 290 |
+
"special": true
|
| 291 |
+
},
|
| 292 |
+
"128036": {
|
| 293 |
+
"content": "<|reserved_special_token_28|>",
|
| 294 |
+
"lstrip": false,
|
| 295 |
+
"normalized": false,
|
| 296 |
+
"rstrip": false,
|
| 297 |
+
"single_word": false,
|
| 298 |
+
"special": true
|
| 299 |
+
},
|
| 300 |
+
"128037": {
|
| 301 |
+
"content": "<|reserved_special_token_29|>",
|
| 302 |
+
"lstrip": false,
|
| 303 |
+
"normalized": false,
|
| 304 |
+
"rstrip": false,
|
| 305 |
+
"single_word": false,
|
| 306 |
+
"special": true
|
| 307 |
+
},
|
| 308 |
+
"128038": {
|
| 309 |
+
"content": "<|reserved_special_token_30|>",
|
| 310 |
+
"lstrip": false,
|
| 311 |
+
"normalized": false,
|
| 312 |
+
"rstrip": false,
|
| 313 |
+
"single_word": false,
|
| 314 |
+
"special": true
|
| 315 |
+
},
|
| 316 |
+
"128039": {
|
| 317 |
+
"content": "<|reserved_special_token_31|>",
|
| 318 |
+
"lstrip": false,
|
| 319 |
+
"normalized": false,
|
| 320 |
+
"rstrip": false,
|
| 321 |
+
"single_word": false,
|
| 322 |
+
"special": true
|
| 323 |
+
},
|
| 324 |
+
"128040": {
|
| 325 |
+
"content": "<|reserved_special_token_32|>",
|
| 326 |
+
"lstrip": false,
|
| 327 |
+
"normalized": false,
|
| 328 |
+
"rstrip": false,
|
| 329 |
+
"single_word": false,
|
| 330 |
+
"special": true
|
| 331 |
+
},
|
| 332 |
+
"128041": {
|
| 333 |
+
"content": "<|reserved_special_token_33|>",
|
| 334 |
+
"lstrip": false,
|
| 335 |
+
"normalized": false,
|
| 336 |
+
"rstrip": false,
|
| 337 |
+
"single_word": false,
|
| 338 |
+
"special": true
|
| 339 |
+
},
|
| 340 |
+
"128042": {
|
| 341 |
+
"content": "<|reserved_special_token_34|>",
|
| 342 |
+
"lstrip": false,
|
| 343 |
+
"normalized": false,
|
| 344 |
+
"rstrip": false,
|
| 345 |
+
"single_word": false,
|
| 346 |
+
"special": true
|
| 347 |
+
},
|
| 348 |
+
"128043": {
|
| 349 |
+
"content": "<|reserved_special_token_35|>",
|
| 350 |
+
"lstrip": false,
|
| 351 |
+
"normalized": false,
|
| 352 |
+
"rstrip": false,
|
| 353 |
+
"single_word": false,
|
| 354 |
+
"special": true
|
| 355 |
+
},
|
| 356 |
+
"128044": {
|
| 357 |
+
"content": "<|reserved_special_token_36|>",
|
| 358 |
+
"lstrip": false,
|
| 359 |
+
"normalized": false,
|
| 360 |
+
"rstrip": false,
|
| 361 |
+
"single_word": false,
|
| 362 |
+
"special": true
|
| 363 |
+
},
|
| 364 |
+
"128045": {
|
| 365 |
+
"content": "<|reserved_special_token_37|>",
|
| 366 |
+
"lstrip": false,
|
| 367 |
+
"normalized": false,
|
| 368 |
+
"rstrip": false,
|
| 369 |
+
"single_word": false,
|
| 370 |
+
"special": true
|
| 371 |
+
},
|
| 372 |
+
"128046": {
|
| 373 |
+
"content": "<|reserved_special_token_38|>",
|
| 374 |
+
"lstrip": false,
|
| 375 |
+
"normalized": false,
|
| 376 |
+
"rstrip": false,
|
| 377 |
+
"single_word": false,
|
| 378 |
+
"special": true
|
| 379 |
+
},
|
| 380 |
+
"128047": {
|
| 381 |
+
"content": "<|reserved_special_token_39|>",
|
| 382 |
+
"lstrip": false,
|
| 383 |
+
"normalized": false,
|
| 384 |
+
"rstrip": false,
|
| 385 |
+
"single_word": false,
|
| 386 |
+
"special": true
|
| 387 |
+
},
|
| 388 |
+
"128048": {
|
| 389 |
+
"content": "<|reserved_special_token_40|>",
|
| 390 |
+
"lstrip": false,
|
| 391 |
+
"normalized": false,
|
| 392 |
+
"rstrip": false,
|
| 393 |
+
"single_word": false,
|
| 394 |
+
"special": true
|
| 395 |
+
},
|
| 396 |
+
"128049": {
|
| 397 |
+
"content": "<|reserved_special_token_41|>",
|
| 398 |
+
"lstrip": false,
|
| 399 |
+
"normalized": false,
|
| 400 |
+
"rstrip": false,
|
| 401 |
+
"single_word": false,
|
| 402 |
+
"special": true
|
| 403 |
+
},
|
| 404 |
+
"128050": {
|
| 405 |
+
"content": "<|reserved_special_token_42|>",
|
| 406 |
+
"lstrip": false,
|
| 407 |
+
"normalized": false,
|
| 408 |
+
"rstrip": false,
|
| 409 |
+
"single_word": false,
|
| 410 |
+
"special": true
|
| 411 |
+
},
|
| 412 |
+
"128051": {
|
| 413 |
+
"content": "<|reserved_special_token_43|>",
|
| 414 |
+
"lstrip": false,
|
| 415 |
+
"normalized": false,
|
| 416 |
+
"rstrip": false,
|
| 417 |
+
"single_word": false,
|
| 418 |
+
"special": true
|
| 419 |
+
},
|
| 420 |
+
"128052": {
|
| 421 |
+
"content": "<|reserved_special_token_44|>",
|
| 422 |
+
"lstrip": false,
|
| 423 |
+
"normalized": false,
|
| 424 |
+
"rstrip": false,
|
| 425 |
+
"single_word": false,
|
| 426 |
+
"special": true
|
| 427 |
+
},
|
| 428 |
+
"128053": {
|
| 429 |
+
"content": "<|reserved_special_token_45|>",
|
| 430 |
+
"lstrip": false,
|
| 431 |
+
"normalized": false,
|
| 432 |
+
"rstrip": false,
|
| 433 |
+
"single_word": false,
|
| 434 |
+
"special": true
|
| 435 |
+
},
|
| 436 |
+
"128054": {
|
| 437 |
+
"content": "<|reserved_special_token_46|>",
|
| 438 |
+
"lstrip": false,
|
| 439 |
+
"normalized": false,
|
| 440 |
+
"rstrip": false,
|
| 441 |
+
"single_word": false,
|
| 442 |
+
"special": true
|
| 443 |
+
},
|
| 444 |
+
"128055": {
|
| 445 |
+
"content": "<|reserved_special_token_47|>",
|
| 446 |
+
"lstrip": false,
|
| 447 |
+
"normalized": false,
|
| 448 |
+
"rstrip": false,
|
| 449 |
+
"single_word": false,
|
| 450 |
+
"special": true
|
| 451 |
+
},
|
| 452 |
+
"128056": {
|
| 453 |
+
"content": "<|reserved_special_token_48|>",
|
| 454 |
+
"lstrip": false,
|
| 455 |
+
"normalized": false,
|
| 456 |
+
"rstrip": false,
|
| 457 |
+
"single_word": false,
|
| 458 |
+
"special": true
|
| 459 |
+
},
|
| 460 |
+
"128057": {
|
| 461 |
+
"content": "<|reserved_special_token_49|>",
|
| 462 |
+
"lstrip": false,
|
| 463 |
+
"normalized": false,
|
| 464 |
+
"rstrip": false,
|
| 465 |
+
"single_word": false,
|
| 466 |
+
"special": true
|
| 467 |
+
},
|
| 468 |
+
"128058": {
|
| 469 |
+
"content": "<|reserved_special_token_50|>",
|
| 470 |
+
"lstrip": false,
|
| 471 |
+
"normalized": false,
|
| 472 |
+
"rstrip": false,
|
| 473 |
+
"single_word": false,
|
| 474 |
+
"special": true
|
| 475 |
+
},
|
| 476 |
+
"128059": {
|
| 477 |
+
"content": "<|reserved_special_token_51|>",
|
| 478 |
+
"lstrip": false,
|
| 479 |
+
"normalized": false,
|
| 480 |
+
"rstrip": false,
|
| 481 |
+
"single_word": false,
|
| 482 |
+
"special": true
|
| 483 |
+
},
|
| 484 |
+
"128060": {
|
| 485 |
+
"content": "<|reserved_special_token_52|>",
|
| 486 |
+
"lstrip": false,
|
| 487 |
+
"normalized": false,
|
| 488 |
+
"rstrip": false,
|
| 489 |
+
"single_word": false,
|
| 490 |
+
"special": true
|
| 491 |
+
},
|
| 492 |
+
"128061": {
|
| 493 |
+
"content": "<|reserved_special_token_53|>",
|
| 494 |
+
"lstrip": false,
|
| 495 |
+
"normalized": false,
|
| 496 |
+
"rstrip": false,
|
| 497 |
+
"single_word": false,
|
| 498 |
+
"special": true
|
| 499 |
+
},
|
| 500 |
+
"128062": {
|
| 501 |
+
"content": "<|reserved_special_token_54|>",
|
| 502 |
+
"lstrip": false,
|
| 503 |
+
"normalized": false,
|
| 504 |
+
"rstrip": false,
|
| 505 |
+
"single_word": false,
|
| 506 |
+
"special": true
|
| 507 |
+
},
|
| 508 |
+
"128063": {
|
| 509 |
+
"content": "<|reserved_special_token_55|>",
|
| 510 |
+
"lstrip": false,
|
| 511 |
+
"normalized": false,
|
| 512 |
+
"rstrip": false,
|
| 513 |
+
"single_word": false,
|
| 514 |
+
"special": true
|
| 515 |
+
},
|
| 516 |
+
"128064": {
|
| 517 |
+
"content": "<|reserved_special_token_56|>",
|
| 518 |
+
"lstrip": false,
|
| 519 |
+
"normalized": false,
|
| 520 |
+
"rstrip": false,
|
| 521 |
+
"single_word": false,
|
| 522 |
+
"special": true
|
| 523 |
+
},
|
| 524 |
+
"128065": {
|
| 525 |
+
"content": "<|reserved_special_token_57|>",
|
| 526 |
+
"lstrip": false,
|
| 527 |
+
"normalized": false,
|
| 528 |
+
"rstrip": false,
|
| 529 |
+
"single_word": false,
|
| 530 |
+
"special": true
|
| 531 |
+
},
|
| 532 |
+
"128066": {
|
| 533 |
+
"content": "<|reserved_special_token_58|>",
|
| 534 |
+
"lstrip": false,
|
| 535 |
+
"normalized": false,
|
| 536 |
+
"rstrip": false,
|
| 537 |
+
"single_word": false,
|
| 538 |
+
"special": true
|
| 539 |
+
},
|
| 540 |
+
"128067": {
|
| 541 |
+
"content": "<|reserved_special_token_59|>",
|
| 542 |
+
"lstrip": false,
|
| 543 |
+
"normalized": false,
|
| 544 |
+
"rstrip": false,
|
| 545 |
+
"single_word": false,
|
| 546 |
+
"special": true
|
| 547 |
+
},
|
| 548 |
+
"128068": {
|
| 549 |
+
"content": "<|reserved_special_token_60|>",
|
| 550 |
+
"lstrip": false,
|
| 551 |
+
"normalized": false,
|
| 552 |
+
"rstrip": false,
|
| 553 |
+
"single_word": false,
|
| 554 |
+
"special": true
|
| 555 |
+
},
|
| 556 |
+
"128069": {
|
| 557 |
+
"content": "<|reserved_special_token_61|>",
|
| 558 |
+
"lstrip": false,
|
| 559 |
+
"normalized": false,
|
| 560 |
+
"rstrip": false,
|
| 561 |
+
"single_word": false,
|
| 562 |
+
"special": true
|
| 563 |
+
},
|
| 564 |
+
"128070": {
|
| 565 |
+
"content": "<|reserved_special_token_62|>",
|
| 566 |
+
"lstrip": false,
|
| 567 |
+
"normalized": false,
|
| 568 |
+
"rstrip": false,
|
| 569 |
+
"single_word": false,
|
| 570 |
+
"special": true
|
| 571 |
+
},
|
| 572 |
+
"128071": {
|
| 573 |
+
"content": "<|reserved_special_token_63|>",
|
| 574 |
+
"lstrip": false,
|
| 575 |
+
"normalized": false,
|
| 576 |
+
"rstrip": false,
|
| 577 |
+
"single_word": false,
|
| 578 |
+
"special": true
|
| 579 |
+
},
|
| 580 |
+
"128072": {
|
| 581 |
+
"content": "<|reserved_special_token_64|>",
|
| 582 |
+
"lstrip": false,
|
| 583 |
+
"normalized": false,
|
| 584 |
+
"rstrip": false,
|
| 585 |
+
"single_word": false,
|
| 586 |
+
"special": true
|
| 587 |
+
},
|
| 588 |
+
"128073": {
|
| 589 |
+
"content": "<|reserved_special_token_65|>",
|
| 590 |
+
"lstrip": false,
|
| 591 |
+
"normalized": false,
|
| 592 |
+
"rstrip": false,
|
| 593 |
+
"single_word": false,
|
| 594 |
+
"special": true
|
| 595 |
+
},
|
| 596 |
+
"128074": {
|
| 597 |
+
"content": "<|reserved_special_token_66|>",
|
| 598 |
+
"lstrip": false,
|
| 599 |
+
"normalized": false,
|
| 600 |
+
"rstrip": false,
|
| 601 |
+
"single_word": false,
|
| 602 |
+
"special": true
|
| 603 |
+
},
|
| 604 |
+
"128075": {
|
| 605 |
+
"content": "<|reserved_special_token_67|>",
|
| 606 |
+
"lstrip": false,
|
| 607 |
+
"normalized": false,
|
| 608 |
+
"rstrip": false,
|
| 609 |
+
"single_word": false,
|
| 610 |
+
"special": true
|
| 611 |
+
},
|
| 612 |
+
"128076": {
|
| 613 |
+
"content": "<|reserved_special_token_68|>",
|
| 614 |
+
"lstrip": false,
|
| 615 |
+
"normalized": false,
|
| 616 |
+
"rstrip": false,
|
| 617 |
+
"single_word": false,
|
| 618 |
+
"special": true
|
| 619 |
+
},
|
| 620 |
+
"128077": {
|
| 621 |
+
"content": "<|reserved_special_token_69|>",
|
| 622 |
+
"lstrip": false,
|
| 623 |
+
"normalized": false,
|
| 624 |
+
"rstrip": false,
|
| 625 |
+
"single_word": false,
|
| 626 |
+
"special": true
|
| 627 |
+
},
|
| 628 |
+
"128078": {
|
| 629 |
+
"content": "<|reserved_special_token_70|>",
|
| 630 |
+
"lstrip": false,
|
| 631 |
+
"normalized": false,
|
| 632 |
+
"rstrip": false,
|
| 633 |
+
"single_word": false,
|
| 634 |
+
"special": true
|
| 635 |
+
},
|
| 636 |
+
"128079": {
|
| 637 |
+
"content": "<|reserved_special_token_71|>",
|
| 638 |
+
"lstrip": false,
|
| 639 |
+
"normalized": false,
|
| 640 |
+
"rstrip": false,
|
| 641 |
+
"single_word": false,
|
| 642 |
+
"special": true
|
| 643 |
+
},
|
| 644 |
+
"128080": {
|
| 645 |
+
"content": "<|reserved_special_token_72|>",
|
| 646 |
+
"lstrip": false,
|
| 647 |
+
"normalized": false,
|
| 648 |
+
"rstrip": false,
|
| 649 |
+
"single_word": false,
|
| 650 |
+
"special": true
|
| 651 |
+
},
|
| 652 |
+
"128081": {
|
| 653 |
+
"content": "<|reserved_special_token_73|>",
|
| 654 |
+
"lstrip": false,
|
| 655 |
+
"normalized": false,
|
| 656 |
+
"rstrip": false,
|
| 657 |
+
"single_word": false,
|
| 658 |
+
"special": true
|
| 659 |
+
},
|
| 660 |
+
"128082": {
|
| 661 |
+
"content": "<|reserved_special_token_74|>",
|
| 662 |
+
"lstrip": false,
|
| 663 |
+
"normalized": false,
|
| 664 |
+
"rstrip": false,
|
| 665 |
+
"single_word": false,
|
| 666 |
+
"special": true
|
| 667 |
+
},
|
| 668 |
+
"128083": {
|
| 669 |
+
"content": "<|reserved_special_token_75|>",
|
| 670 |
+
"lstrip": false,
|
| 671 |
+
"normalized": false,
|
| 672 |
+
"rstrip": false,
|
| 673 |
+
"single_word": false,
|
| 674 |
+
"special": true
|
| 675 |
+
},
|
| 676 |
+
"128084": {
|
| 677 |
+
"content": "<|reserved_special_token_76|>",
|
| 678 |
+
"lstrip": false,
|
| 679 |
+
"normalized": false,
|
| 680 |
+
"rstrip": false,
|
| 681 |
+
"single_word": false,
|
| 682 |
+
"special": true
|
| 683 |
+
},
|
| 684 |
+
"128085": {
|
| 685 |
+
"content": "<|reserved_special_token_77|>",
|
| 686 |
+
"lstrip": false,
|
| 687 |
+
"normalized": false,
|
| 688 |
+
"rstrip": false,
|
| 689 |
+
"single_word": false,
|
| 690 |
+
"special": true
|
| 691 |
+
},
|
| 692 |
+
"128086": {
|
| 693 |
+
"content": "<|reserved_special_token_78|>",
|
| 694 |
+
"lstrip": false,
|
| 695 |
+
"normalized": false,
|
| 696 |
+
"rstrip": false,
|
| 697 |
+
"single_word": false,
|
| 698 |
+
"special": true
|
| 699 |
+
},
|
| 700 |
+
"128087": {
|
| 701 |
+
"content": "<|reserved_special_token_79|>",
|
| 702 |
+
"lstrip": false,
|
| 703 |
+
"normalized": false,
|
| 704 |
+
"rstrip": false,
|
| 705 |
+
"single_word": false,
|
| 706 |
+
"special": true
|
| 707 |
+
},
|
| 708 |
+
"128088": {
|
| 709 |
+
"content": "<|reserved_special_token_80|>",
|
| 710 |
+
"lstrip": false,
|
| 711 |
+
"normalized": false,
|
| 712 |
+
"rstrip": false,
|
| 713 |
+
"single_word": false,
|
| 714 |
+
"special": true
|
| 715 |
+
},
|
| 716 |
+
"128089": {
|
| 717 |
+
"content": "<|reserved_special_token_81|>",
|
| 718 |
+
"lstrip": false,
|
| 719 |
+
"normalized": false,
|
| 720 |
+
"rstrip": false,
|
| 721 |
+
"single_word": false,
|
| 722 |
+
"special": true
|
| 723 |
+
},
|
| 724 |
+
"128090": {
|
| 725 |
+
"content": "<|reserved_special_token_82|>",
|
| 726 |
+
"lstrip": false,
|
| 727 |
+
"normalized": false,
|
| 728 |
+
"rstrip": false,
|
| 729 |
+
"single_word": false,
|
| 730 |
+
"special": true
|
| 731 |
+
},
|
| 732 |
+
"128091": {
|
| 733 |
+
"content": "<|reserved_special_token_83|>",
|
| 734 |
+
"lstrip": false,
|
| 735 |
+
"normalized": false,
|
| 736 |
+
"rstrip": false,
|
| 737 |
+
"single_word": false,
|
| 738 |
+
"special": true
|
| 739 |
+
},
|
| 740 |
+
"128092": {
|
| 741 |
+
"content": "<|reserved_special_token_84|>",
|
| 742 |
+
"lstrip": false,
|
| 743 |
+
"normalized": false,
|
| 744 |
+
"rstrip": false,
|
| 745 |
+
"single_word": false,
|
| 746 |
+
"special": true
|
| 747 |
+
},
|
| 748 |
+
"128093": {
|
| 749 |
+
"content": "<|reserved_special_token_85|>",
|
| 750 |
+
"lstrip": false,
|
| 751 |
+
"normalized": false,
|
| 752 |
+
"rstrip": false,
|
| 753 |
+
"single_word": false,
|
| 754 |
+
"special": true
|
| 755 |
+
},
|
| 756 |
+
"128094": {
|
| 757 |
+
"content": "<|reserved_special_token_86|>",
|
| 758 |
+
"lstrip": false,
|
| 759 |
+
"normalized": false,
|
| 760 |
+
"rstrip": false,
|
| 761 |
+
"single_word": false,
|
| 762 |
+
"special": true
|
| 763 |
+
},
|
| 764 |
+
"128095": {
|
| 765 |
+
"content": "<|reserved_special_token_87|>",
|
| 766 |
+
"lstrip": false,
|
| 767 |
+
"normalized": false,
|
| 768 |
+
"rstrip": false,
|
| 769 |
+
"single_word": false,
|
| 770 |
+
"special": true
|
| 771 |
+
},
|
| 772 |
+
"128096": {
|
| 773 |
+
"content": "<|reserved_special_token_88|>",
|
| 774 |
+
"lstrip": false,
|
| 775 |
+
"normalized": false,
|
| 776 |
+
"rstrip": false,
|
| 777 |
+
"single_word": false,
|
| 778 |
+
"special": true
|
| 779 |
+
},
|
| 780 |
+
"128097": {
|
| 781 |
+
"content": "<|reserved_special_token_89|>",
|
| 782 |
+
"lstrip": false,
|
| 783 |
+
"normalized": false,
|
| 784 |
+
"rstrip": false,
|
| 785 |
+
"single_word": false,
|
| 786 |
+
"special": true
|
| 787 |
+
},
|
| 788 |
+
"128098": {
|
| 789 |
+
"content": "<|reserved_special_token_90|>",
|
| 790 |
+
"lstrip": false,
|
| 791 |
+
"normalized": false,
|
| 792 |
+
"rstrip": false,
|
| 793 |
+
"single_word": false,
|
| 794 |
+
"special": true
|
| 795 |
+
},
|
| 796 |
+
"128099": {
|
| 797 |
+
"content": "<|reserved_special_token_91|>",
|
| 798 |
+
"lstrip": false,
|
| 799 |
+
"normalized": false,
|
| 800 |
+
"rstrip": false,
|
| 801 |
+
"single_word": false,
|
| 802 |
+
"special": true
|
| 803 |
+
},
|
| 804 |
+
"128100": {
|
| 805 |
+
"content": "<|reserved_special_token_92|>",
|
| 806 |
+
"lstrip": false,
|
| 807 |
+
"normalized": false,
|
| 808 |
+
"rstrip": false,
|
| 809 |
+
"single_word": false,
|
| 810 |
+
"special": true
|
| 811 |
+
},
|
| 812 |
+
"128101": {
|
| 813 |
+
"content": "<|reserved_special_token_93|>",
|
| 814 |
+
"lstrip": false,
|
| 815 |
+
"normalized": false,
|
| 816 |
+
"rstrip": false,
|
| 817 |
+
"single_word": false,
|
| 818 |
+
"special": true
|
| 819 |
+
},
|
| 820 |
+
"128102": {
|
| 821 |
+
"content": "<|reserved_special_token_94|>",
|
| 822 |
+
"lstrip": false,
|
| 823 |
+
"normalized": false,
|
| 824 |
+
"rstrip": false,
|
| 825 |
+
"single_word": false,
|
| 826 |
+
"special": true
|
| 827 |
+
},
|
| 828 |
+
"128103": {
|
| 829 |
+
"content": "<|reserved_special_token_95|>",
|
| 830 |
+
"lstrip": false,
|
| 831 |
+
"normalized": false,
|
| 832 |
+
"rstrip": false,
|
| 833 |
+
"single_word": false,
|
| 834 |
+
"special": true
|
| 835 |
+
},
|
| 836 |
+
"128104": {
|
| 837 |
+
"content": "<|reserved_special_token_96|>",
|
| 838 |
+
"lstrip": false,
|
| 839 |
+
"normalized": false,
|
| 840 |
+
"rstrip": false,
|
| 841 |
+
"single_word": false,
|
| 842 |
+
"special": true
|
| 843 |
+
},
|
| 844 |
+
"128105": {
|
| 845 |
+
"content": "<|reserved_special_token_97|>",
|
| 846 |
+
"lstrip": false,
|
| 847 |
+
"normalized": false,
|
| 848 |
+
"rstrip": false,
|
| 849 |
+
"single_word": false,
|
| 850 |
+
"special": true
|
| 851 |
+
},
|
| 852 |
+
"128106": {
|
| 853 |
+
"content": "<|reserved_special_token_98|>",
|
| 854 |
+
"lstrip": false,
|
| 855 |
+
"normalized": false,
|
| 856 |
+
"rstrip": false,
|
| 857 |
+
"single_word": false,
|
| 858 |
+
"special": true
|
| 859 |
+
},
|
| 860 |
+
"128107": {
|
| 861 |
+
"content": "<|reserved_special_token_99|>",
|
| 862 |
+
"lstrip": false,
|
| 863 |
+
"normalized": false,
|
| 864 |
+
"rstrip": false,
|
| 865 |
+
"single_word": false,
|
| 866 |
+
"special": true
|
| 867 |
+
},
|
| 868 |
+
"128108": {
|
| 869 |
+
"content": "<|reserved_special_token_100|>",
|
| 870 |
+
"lstrip": false,
|
| 871 |
+
"normalized": false,
|
| 872 |
+
"rstrip": false,
|
| 873 |
+
"single_word": false,
|
| 874 |
+
"special": true
|
| 875 |
+
},
|
| 876 |
+
"128109": {
|
| 877 |
+
"content": "<|reserved_special_token_101|>",
|
| 878 |
+
"lstrip": false,
|
| 879 |
+
"normalized": false,
|
| 880 |
+
"rstrip": false,
|
| 881 |
+
"single_word": false,
|
| 882 |
+
"special": true
|
| 883 |
+
},
|
| 884 |
+
"128110": {
|
| 885 |
+
"content": "<|reserved_special_token_102|>",
|
| 886 |
+
"lstrip": false,
|
| 887 |
+
"normalized": false,
|
| 888 |
+
"rstrip": false,
|
| 889 |
+
"single_word": false,
|
| 890 |
+
"special": true
|
| 891 |
+
},
|
| 892 |
+
"128111": {
|
| 893 |
+
"content": "<|reserved_special_token_103|>",
|
| 894 |
+
"lstrip": false,
|
| 895 |
+
"normalized": false,
|
| 896 |
+
"rstrip": false,
|
| 897 |
+
"single_word": false,
|
| 898 |
+
"special": true
|
| 899 |
+
},
|
| 900 |
+
"128112": {
|
| 901 |
+
"content": "<|reserved_special_token_104|>",
|
| 902 |
+
"lstrip": false,
|
| 903 |
+
"normalized": false,
|
| 904 |
+
"rstrip": false,
|
| 905 |
+
"single_word": false,
|
| 906 |
+
"special": true
|
| 907 |
+
},
|
| 908 |
+
"128113": {
|
| 909 |
+
"content": "<|reserved_special_token_105|>",
|
| 910 |
+
"lstrip": false,
|
| 911 |
+
"normalized": false,
|
| 912 |
+
"rstrip": false,
|
| 913 |
+
"single_word": false,
|
| 914 |
+
"special": true
|
| 915 |
+
},
|
| 916 |
+
"128114": {
|
| 917 |
+
"content": "<|reserved_special_token_106|>",
|
| 918 |
+
"lstrip": false,
|
| 919 |
+
"normalized": false,
|
| 920 |
+
"rstrip": false,
|
| 921 |
+
"single_word": false,
|
| 922 |
+
"special": true
|
| 923 |
+
},
|
| 924 |
+
"128115": {
|
| 925 |
+
"content": "<|reserved_special_token_107|>",
|
| 926 |
+
"lstrip": false,
|
| 927 |
+
"normalized": false,
|
| 928 |
+
"rstrip": false,
|
| 929 |
+
"single_word": false,
|
| 930 |
+
"special": true
|
| 931 |
+
},
|
| 932 |
+
"128116": {
|
| 933 |
+
"content": "<|reserved_special_token_108|>",
|
| 934 |
+
"lstrip": false,
|
| 935 |
+
"normalized": false,
|
| 936 |
+
"rstrip": false,
|
| 937 |
+
"single_word": false,
|
| 938 |
+
"special": true
|
| 939 |
+
},
|
| 940 |
+
"128117": {
|
| 941 |
+
"content": "<|reserved_special_token_109|>",
|
| 942 |
+
"lstrip": false,
|
| 943 |
+
"normalized": false,
|
| 944 |
+
"rstrip": false,
|
| 945 |
+
"single_word": false,
|
| 946 |
+
"special": true
|
| 947 |
+
},
|
| 948 |
+
"128118": {
|
| 949 |
+
"content": "<|reserved_special_token_110|>",
|
| 950 |
+
"lstrip": false,
|
| 951 |
+
"normalized": false,
|
| 952 |
+
"rstrip": false,
|
| 953 |
+
"single_word": false,
|
| 954 |
+
"special": true
|
| 955 |
+
},
|
| 956 |
+
"128119": {
|
| 957 |
+
"content": "<|reserved_special_token_111|>",
|
| 958 |
+
"lstrip": false,
|
| 959 |
+
"normalized": false,
|
| 960 |
+
"rstrip": false,
|
| 961 |
+
"single_word": false,
|
| 962 |
+
"special": true
|
| 963 |
+
},
|
| 964 |
+
"128120": {
|
| 965 |
+
"content": "<|reserved_special_token_112|>",
|
| 966 |
+
"lstrip": false,
|
| 967 |
+
"normalized": false,
|
| 968 |
+
"rstrip": false,
|
| 969 |
+
"single_word": false,
|
| 970 |
+
"special": true
|
| 971 |
+
},
|
| 972 |
+
"128121": {
|
| 973 |
+
"content": "<|reserved_special_token_113|>",
|
| 974 |
+
"lstrip": false,
|
| 975 |
+
"normalized": false,
|
| 976 |
+
"rstrip": false,
|
| 977 |
+
"single_word": false,
|
| 978 |
+
"special": true
|
| 979 |
+
},
|
| 980 |
+
"128122": {
|
| 981 |
+
"content": "<|reserved_special_token_114|>",
|
| 982 |
+
"lstrip": false,
|
| 983 |
+
"normalized": false,
|
| 984 |
+
"rstrip": false,
|
| 985 |
+
"single_word": false,
|
| 986 |
+
"special": true
|
| 987 |
+
},
|
| 988 |
+
"128123": {
|
| 989 |
+
"content": "<|reserved_special_token_115|>",
|
| 990 |
+
"lstrip": false,
|
| 991 |
+
"normalized": false,
|
| 992 |
+
"rstrip": false,
|
| 993 |
+
"single_word": false,
|
| 994 |
+
"special": true
|
| 995 |
+
},
|
| 996 |
+
"128124": {
|
| 997 |
+
"content": "<|reserved_special_token_116|>",
|
| 998 |
+
"lstrip": false,
|
| 999 |
+
"normalized": false,
|
| 1000 |
+
"rstrip": false,
|
| 1001 |
+
"single_word": false,
|
| 1002 |
+
"special": true
|
| 1003 |
+
},
|
| 1004 |
+
"128125": {
|
| 1005 |
+
"content": "<|reserved_special_token_117|>",
|
| 1006 |
+
"lstrip": false,
|
| 1007 |
+
"normalized": false,
|
| 1008 |
+
"rstrip": false,
|
| 1009 |
+
"single_word": false,
|
| 1010 |
+
"special": true
|
| 1011 |
+
},
|
| 1012 |
+
"128126": {
|
| 1013 |
+
"content": "<|reserved_special_token_118|>",
|
| 1014 |
+
"lstrip": false,
|
| 1015 |
+
"normalized": false,
|
| 1016 |
+
"rstrip": false,
|
| 1017 |
+
"single_word": false,
|
| 1018 |
+
"special": true
|
| 1019 |
+
},
|
| 1020 |
+
"128127": {
|
| 1021 |
+
"content": "<|reserved_special_token_119|>",
|
| 1022 |
+
"lstrip": false,
|
| 1023 |
+
"normalized": false,
|
| 1024 |
+
"rstrip": false,
|
| 1025 |
+
"single_word": false,
|
| 1026 |
+
"special": true
|
| 1027 |
+
},
|
| 1028 |
+
"128128": {
|
| 1029 |
+
"content": "<|reserved_special_token_120|>",
|
| 1030 |
+
"lstrip": false,
|
| 1031 |
+
"normalized": false,
|
| 1032 |
+
"rstrip": false,
|
| 1033 |
+
"single_word": false,
|
| 1034 |
+
"special": true
|
| 1035 |
+
},
|
| 1036 |
+
"128129": {
|
| 1037 |
+
"content": "<|reserved_special_token_121|>",
|
| 1038 |
+
"lstrip": false,
|
| 1039 |
+
"normalized": false,
|
| 1040 |
+
"rstrip": false,
|
| 1041 |
+
"single_word": false,
|
| 1042 |
+
"special": true
|
| 1043 |
+
},
|
| 1044 |
+
"128130": {
|
| 1045 |
+
"content": "<|reserved_special_token_122|>",
|
| 1046 |
+
"lstrip": false,
|
| 1047 |
+
"normalized": false,
|
| 1048 |
+
"rstrip": false,
|
| 1049 |
+
"single_word": false,
|
| 1050 |
+
"special": true
|
| 1051 |
+
},
|
| 1052 |
+
"128131": {
|
| 1053 |
+
"content": "<|reserved_special_token_123|>",
|
| 1054 |
+
"lstrip": false,
|
| 1055 |
+
"normalized": false,
|
| 1056 |
+
"rstrip": false,
|
| 1057 |
+
"single_word": false,
|
| 1058 |
+
"special": true
|
| 1059 |
+
},
|
| 1060 |
+
"128132": {
|
| 1061 |
+
"content": "<|reserved_special_token_124|>",
|
| 1062 |
+
"lstrip": false,
|
| 1063 |
+
"normalized": false,
|
| 1064 |
+
"rstrip": false,
|
| 1065 |
+
"single_word": false,
|
| 1066 |
+
"special": true
|
| 1067 |
+
},
|
| 1068 |
+
"128133": {
|
| 1069 |
+
"content": "<|reserved_special_token_125|>",
|
| 1070 |
+
"lstrip": false,
|
| 1071 |
+
"normalized": false,
|
| 1072 |
+
"rstrip": false,
|
| 1073 |
+
"single_word": false,
|
| 1074 |
+
"special": true
|
| 1075 |
+
},
|
| 1076 |
+
"128134": {
|
| 1077 |
+
"content": "<|reserved_special_token_126|>",
|
| 1078 |
+
"lstrip": false,
|
| 1079 |
+
"normalized": false,
|
| 1080 |
+
"rstrip": false,
|
| 1081 |
+
"single_word": false,
|
| 1082 |
+
"special": true
|
| 1083 |
+
},
|
| 1084 |
+
"128135": {
|
| 1085 |
+
"content": "<|reserved_special_token_127|>",
|
| 1086 |
+
"lstrip": false,
|
| 1087 |
+
"normalized": false,
|
| 1088 |
+
"rstrip": false,
|
| 1089 |
+
"single_word": false,
|
| 1090 |
+
"special": true
|
| 1091 |
+
},
|
| 1092 |
+
"128136": {
|
| 1093 |
+
"content": "<|reserved_special_token_128|>",
|
| 1094 |
+
"lstrip": false,
|
| 1095 |
+
"normalized": false,
|
| 1096 |
+
"rstrip": false,
|
| 1097 |
+
"single_word": false,
|
| 1098 |
+
"special": true
|
| 1099 |
+
},
|
| 1100 |
+
"128137": {
|
| 1101 |
+
"content": "<|reserved_special_token_129|>",
|
| 1102 |
+
"lstrip": false,
|
| 1103 |
+
"normalized": false,
|
| 1104 |
+
"rstrip": false,
|
| 1105 |
+
"single_word": false,
|
| 1106 |
+
"special": true
|
| 1107 |
+
},
|
| 1108 |
+
"128138": {
|
| 1109 |
+
"content": "<|reserved_special_token_130|>",
|
| 1110 |
+
"lstrip": false,
|
| 1111 |
+
"normalized": false,
|
| 1112 |
+
"rstrip": false,
|
| 1113 |
+
"single_word": false,
|
| 1114 |
+
"special": true
|
| 1115 |
+
},
|
| 1116 |
+
"128139": {
|
| 1117 |
+
"content": "<|reserved_special_token_131|>",
|
| 1118 |
+
"lstrip": false,
|
| 1119 |
+
"normalized": false,
|
| 1120 |
+
"rstrip": false,
|
| 1121 |
+
"single_word": false,
|
| 1122 |
+
"special": true
|
| 1123 |
+
},
|
| 1124 |
+
"128140": {
|
| 1125 |
+
"content": "<|reserved_special_token_132|>",
|
| 1126 |
+
"lstrip": false,
|
| 1127 |
+
"normalized": false,
|
| 1128 |
+
"rstrip": false,
|
| 1129 |
+
"single_word": false,
|
| 1130 |
+
"special": true
|
| 1131 |
+
},
|
| 1132 |
+
"128141": {
|
| 1133 |
+
"content": "<|reserved_special_token_133|>",
|
| 1134 |
+
"lstrip": false,
|
| 1135 |
+
"normalized": false,
|
| 1136 |
+
"rstrip": false,
|
| 1137 |
+
"single_word": false,
|
| 1138 |
+
"special": true
|
| 1139 |
+
},
|
| 1140 |
+
"128142": {
|
| 1141 |
+
"content": "<|reserved_special_token_134|>",
|
| 1142 |
+
"lstrip": false,
|
| 1143 |
+
"normalized": false,
|
| 1144 |
+
"rstrip": false,
|
| 1145 |
+
"single_word": false,
|
| 1146 |
+
"special": true
|
| 1147 |
+
},
|
| 1148 |
+
"128143": {
|
| 1149 |
+
"content": "<|reserved_special_token_135|>",
|
| 1150 |
+
"lstrip": false,
|
| 1151 |
+
"normalized": false,
|
| 1152 |
+
"rstrip": false,
|
| 1153 |
+
"single_word": false,
|
| 1154 |
+
"special": true
|
| 1155 |
+
},
|
| 1156 |
+
"128144": {
|
| 1157 |
+
"content": "<|reserved_special_token_136|>",
|
| 1158 |
+
"lstrip": false,
|
| 1159 |
+
"normalized": false,
|
| 1160 |
+
"rstrip": false,
|
| 1161 |
+
"single_word": false,
|
| 1162 |
+
"special": true
|
| 1163 |
+
},
|
| 1164 |
+
"128145": {
|
| 1165 |
+
"content": "<|reserved_special_token_137|>",
|
| 1166 |
+
"lstrip": false,
|
| 1167 |
+
"normalized": false,
|
| 1168 |
+
"rstrip": false,
|
| 1169 |
+
"single_word": false,
|
| 1170 |
+
"special": true
|
| 1171 |
+
},
|
| 1172 |
+
"128146": {
|
| 1173 |
+
"content": "<|reserved_special_token_138|>",
|
| 1174 |
+
"lstrip": false,
|
| 1175 |
+
"normalized": false,
|
| 1176 |
+
"rstrip": false,
|
| 1177 |
+
"single_word": false,
|
| 1178 |
+
"special": true
|
| 1179 |
+
},
|
| 1180 |
+
"128147": {
|
| 1181 |
+
"content": "<|reserved_special_token_139|>",
|
| 1182 |
+
"lstrip": false,
|
| 1183 |
+
"normalized": false,
|
| 1184 |
+
"rstrip": false,
|
| 1185 |
+
"single_word": false,
|
| 1186 |
+
"special": true
|
| 1187 |
+
},
|
| 1188 |
+
"128148": {
|
| 1189 |
+
"content": "<|reserved_special_token_140|>",
|
| 1190 |
+
"lstrip": false,
|
| 1191 |
+
"normalized": false,
|
| 1192 |
+
"rstrip": false,
|
| 1193 |
+
"single_word": false,
|
| 1194 |
+
"special": true
|
| 1195 |
+
},
|
| 1196 |
+
"128149": {
|
| 1197 |
+
"content": "<|reserved_special_token_141|>",
|
| 1198 |
+
"lstrip": false,
|
| 1199 |
+
"normalized": false,
|
| 1200 |
+
"rstrip": false,
|
| 1201 |
+
"single_word": false,
|
| 1202 |
+
"special": true
|
| 1203 |
+
},
|
| 1204 |
+
"128150": {
|
| 1205 |
+
"content": "<|reserved_special_token_142|>",
|
| 1206 |
+
"lstrip": false,
|
| 1207 |
+
"normalized": false,
|
| 1208 |
+
"rstrip": false,
|
| 1209 |
+
"single_word": false,
|
| 1210 |
+
"special": true
|
| 1211 |
+
},
|
| 1212 |
+
"128151": {
|
| 1213 |
+
"content": "<|reserved_special_token_143|>",
|
| 1214 |
+
"lstrip": false,
|
| 1215 |
+
"normalized": false,
|
| 1216 |
+
"rstrip": false,
|
| 1217 |
+
"single_word": false,
|
| 1218 |
+
"special": true
|
| 1219 |
+
},
|
| 1220 |
+
"128152": {
|
| 1221 |
+
"content": "<|reserved_special_token_144|>",
|
| 1222 |
+
"lstrip": false,
|
| 1223 |
+
"normalized": false,
|
| 1224 |
+
"rstrip": false,
|
| 1225 |
+
"single_word": false,
|
| 1226 |
+
"special": true
|
| 1227 |
+
},
|
| 1228 |
+
"128153": {
|
| 1229 |
+
"content": "<|reserved_special_token_145|>",
|
| 1230 |
+
"lstrip": false,
|
| 1231 |
+
"normalized": false,
|
| 1232 |
+
"rstrip": false,
|
| 1233 |
+
"single_word": false,
|
| 1234 |
+
"special": true
|
| 1235 |
+
},
|
| 1236 |
+
"128154": {
|
| 1237 |
+
"content": "<|reserved_special_token_146|>",
|
| 1238 |
+
"lstrip": false,
|
| 1239 |
+
"normalized": false,
|
| 1240 |
+
"rstrip": false,
|
| 1241 |
+
"single_word": false,
|
| 1242 |
+
"special": true
|
| 1243 |
+
},
|
| 1244 |
+
"128155": {
|
| 1245 |
+
"content": "<|reserved_special_token_147|>",
|
| 1246 |
+
"lstrip": false,
|
| 1247 |
+
"normalized": false,
|
| 1248 |
+
"rstrip": false,
|
| 1249 |
+
"single_word": false,
|
| 1250 |
+
"special": true
|
| 1251 |
+
},
|
| 1252 |
+
"128156": {
|
| 1253 |
+
"content": "<|reserved_special_token_148|>",
|
| 1254 |
+
"lstrip": false,
|
| 1255 |
+
"normalized": false,
|
| 1256 |
+
"rstrip": false,
|
| 1257 |
+
"single_word": false,
|
| 1258 |
+
"special": true
|
| 1259 |
+
},
|
| 1260 |
+
"128157": {
|
| 1261 |
+
"content": "<|reserved_special_token_149|>",
|
| 1262 |
+
"lstrip": false,
|
| 1263 |
+
"normalized": false,
|
| 1264 |
+
"rstrip": false,
|
| 1265 |
+
"single_word": false,
|
| 1266 |
+
"special": true
|
| 1267 |
+
},
|
| 1268 |
+
"128158": {
|
| 1269 |
+
"content": "<|reserved_special_token_150|>",
|
| 1270 |
+
"lstrip": false,
|
| 1271 |
+
"normalized": false,
|
| 1272 |
+
"rstrip": false,
|
| 1273 |
+
"single_word": false,
|
| 1274 |
+
"special": true
|
| 1275 |
+
},
|
| 1276 |
+
"128159": {
|
| 1277 |
+
"content": "<|reserved_special_token_151|>",
|
| 1278 |
+
"lstrip": false,
|
| 1279 |
+
"normalized": false,
|
| 1280 |
+
"rstrip": false,
|
| 1281 |
+
"single_word": false,
|
| 1282 |
+
"special": true
|
| 1283 |
+
},
|
| 1284 |
+
"128160": {
|
| 1285 |
+
"content": "<|reserved_special_token_152|>",
|
| 1286 |
+
"lstrip": false,
|
| 1287 |
+
"normalized": false,
|
| 1288 |
+
"rstrip": false,
|
| 1289 |
+
"single_word": false,
|
| 1290 |
+
"special": true
|
| 1291 |
+
},
|
| 1292 |
+
"128161": {
|
| 1293 |
+
"content": "<|reserved_special_token_153|>",
|
| 1294 |
+
"lstrip": false,
|
| 1295 |
+
"normalized": false,
|
| 1296 |
+
"rstrip": false,
|
| 1297 |
+
"single_word": false,
|
| 1298 |
+
"special": true
|
| 1299 |
+
},
|
| 1300 |
+
"128162": {
|
| 1301 |
+
"content": "<|reserved_special_token_154|>",
|
| 1302 |
+
"lstrip": false,
|
| 1303 |
+
"normalized": false,
|
| 1304 |
+
"rstrip": false,
|
| 1305 |
+
"single_word": false,
|
| 1306 |
+
"special": true
|
| 1307 |
+
},
|
| 1308 |
+
"128163": {
|
| 1309 |
+
"content": "<|reserved_special_token_155|>",
|
| 1310 |
+
"lstrip": false,
|
| 1311 |
+
"normalized": false,
|
| 1312 |
+
"rstrip": false,
|
| 1313 |
+
"single_word": false,
|
| 1314 |
+
"special": true
|
| 1315 |
+
},
|
| 1316 |
+
"128164": {
|
| 1317 |
+
"content": "<|reserved_special_token_156|>",
|
| 1318 |
+
"lstrip": false,
|
| 1319 |
+
"normalized": false,
|
| 1320 |
+
"rstrip": false,
|
| 1321 |
+
"single_word": false,
|
| 1322 |
+
"special": true
|
| 1323 |
+
},
|
| 1324 |
+
"128165": {
|
| 1325 |
+
"content": "<|reserved_special_token_157|>",
|
| 1326 |
+
"lstrip": false,
|
| 1327 |
+
"normalized": false,
|
| 1328 |
+
"rstrip": false,
|
| 1329 |
+
"single_word": false,
|
| 1330 |
+
"special": true
|
| 1331 |
+
},
|
| 1332 |
+
"128166": {
|
| 1333 |
+
"content": "<|reserved_special_token_158|>",
|
| 1334 |
+
"lstrip": false,
|
| 1335 |
+
"normalized": false,
|
| 1336 |
+
"rstrip": false,
|
| 1337 |
+
"single_word": false,
|
| 1338 |
+
"special": true
|
| 1339 |
+
},
|
| 1340 |
+
"128167": {
|
| 1341 |
+
"content": "<|reserved_special_token_159|>",
|
| 1342 |
+
"lstrip": false,
|
| 1343 |
+
"normalized": false,
|
| 1344 |
+
"rstrip": false,
|
| 1345 |
+
"single_word": false,
|
| 1346 |
+
"special": true
|
| 1347 |
+
},
|
| 1348 |
+
"128168": {
|
| 1349 |
+
"content": "<|reserved_special_token_160|>",
|
| 1350 |
+
"lstrip": false,
|
| 1351 |
+
"normalized": false,
|
| 1352 |
+
"rstrip": false,
|
| 1353 |
+
"single_word": false,
|
| 1354 |
+
"special": true
|
| 1355 |
+
},
|
| 1356 |
+
"128169": {
|
| 1357 |
+
"content": "<|reserved_special_token_161|>",
|
| 1358 |
+
"lstrip": false,
|
| 1359 |
+
"normalized": false,
|
| 1360 |
+
"rstrip": false,
|
| 1361 |
+
"single_word": false,
|
| 1362 |
+
"special": true
|
| 1363 |
+
},
|
| 1364 |
+
"128170": {
|
| 1365 |
+
"content": "<|reserved_special_token_162|>",
|
| 1366 |
+
"lstrip": false,
|
| 1367 |
+
"normalized": false,
|
| 1368 |
+
"rstrip": false,
|
| 1369 |
+
"single_word": false,
|
| 1370 |
+
"special": true
|
| 1371 |
+
},
|
| 1372 |
+
"128171": {
|
| 1373 |
+
"content": "<|reserved_special_token_163|>",
|
| 1374 |
+
"lstrip": false,
|
| 1375 |
+
"normalized": false,
|
| 1376 |
+
"rstrip": false,
|
| 1377 |
+
"single_word": false,
|
| 1378 |
+
"special": true
|
| 1379 |
+
},
|
| 1380 |
+
"128172": {
|
| 1381 |
+
"content": "<|reserved_special_token_164|>",
|
| 1382 |
+
"lstrip": false,
|
| 1383 |
+
"normalized": false,
|
| 1384 |
+
"rstrip": false,
|
| 1385 |
+
"single_word": false,
|
| 1386 |
+
"special": true
|
| 1387 |
+
},
|
| 1388 |
+
"128173": {
|
| 1389 |
+
"content": "<|reserved_special_token_165|>",
|
| 1390 |
+
"lstrip": false,
|
| 1391 |
+
"normalized": false,
|
| 1392 |
+
"rstrip": false,
|
| 1393 |
+
"single_word": false,
|
| 1394 |
+
"special": true
|
| 1395 |
+
},
|
| 1396 |
+
"128174": {
|
| 1397 |
+
"content": "<|reserved_special_token_166|>",
|
| 1398 |
+
"lstrip": false,
|
| 1399 |
+
"normalized": false,
|
| 1400 |
+
"rstrip": false,
|
| 1401 |
+
"single_word": false,
|
| 1402 |
+
"special": true
|
| 1403 |
+
},
|
| 1404 |
+
"128175": {
|
| 1405 |
+
"content": "<|reserved_special_token_167|>",
|
| 1406 |
+
"lstrip": false,
|
| 1407 |
+
"normalized": false,
|
| 1408 |
+
"rstrip": false,
|
| 1409 |
+
"single_word": false,
|
| 1410 |
+
"special": true
|
| 1411 |
+
},
|
| 1412 |
+
"128176": {
|
| 1413 |
+
"content": "<|reserved_special_token_168|>",
|
| 1414 |
+
"lstrip": false,
|
| 1415 |
+
"normalized": false,
|
| 1416 |
+
"rstrip": false,
|
| 1417 |
+
"single_word": false,
|
| 1418 |
+
"special": true
|
| 1419 |
+
},
|
| 1420 |
+
"128177": {
|
| 1421 |
+
"content": "<|reserved_special_token_169|>",
|
| 1422 |
+
"lstrip": false,
|
| 1423 |
+
"normalized": false,
|
| 1424 |
+
"rstrip": false,
|
| 1425 |
+
"single_word": false,
|
| 1426 |
+
"special": true
|
| 1427 |
+
},
|
| 1428 |
+
"128178": {
|
| 1429 |
+
"content": "<|reserved_special_token_170|>",
|
| 1430 |
+
"lstrip": false,
|
| 1431 |
+
"normalized": false,
|
| 1432 |
+
"rstrip": false,
|
| 1433 |
+
"single_word": false,
|
| 1434 |
+
"special": true
|
| 1435 |
+
},
|
| 1436 |
+
"128179": {
|
| 1437 |
+
"content": "<|reserved_special_token_171|>",
|
| 1438 |
+
"lstrip": false,
|
| 1439 |
+
"normalized": false,
|
| 1440 |
+
"rstrip": false,
|
| 1441 |
+
"single_word": false,
|
| 1442 |
+
"special": true
|
| 1443 |
+
},
|
| 1444 |
+
"128180": {
|
| 1445 |
+
"content": "<|reserved_special_token_172|>",
|
| 1446 |
+
"lstrip": false,
|
| 1447 |
+
"normalized": false,
|
| 1448 |
+
"rstrip": false,
|
| 1449 |
+
"single_word": false,
|
| 1450 |
+
"special": true
|
| 1451 |
+
},
|
| 1452 |
+
"128181": {
|
| 1453 |
+
"content": "<|reserved_special_token_173|>",
|
| 1454 |
+
"lstrip": false,
|
| 1455 |
+
"normalized": false,
|
| 1456 |
+
"rstrip": false,
|
| 1457 |
+
"single_word": false,
|
| 1458 |
+
"special": true
|
| 1459 |
+
},
|
| 1460 |
+
"128182": {
|
| 1461 |
+
"content": "<|reserved_special_token_174|>",
|
| 1462 |
+
"lstrip": false,
|
| 1463 |
+
"normalized": false,
|
| 1464 |
+
"rstrip": false,
|
| 1465 |
+
"single_word": false,
|
| 1466 |
+
"special": true
|
| 1467 |
+
},
|
| 1468 |
+
"128183": {
|
| 1469 |
+
"content": "<|reserved_special_token_175|>",
|
| 1470 |
+
"lstrip": false,
|
| 1471 |
+
"normalized": false,
|
| 1472 |
+
"rstrip": false,
|
| 1473 |
+
"single_word": false,
|
| 1474 |
+
"special": true
|
| 1475 |
+
},
|
| 1476 |
+
"128184": {
|
| 1477 |
+
"content": "<|reserved_special_token_176|>",
|
| 1478 |
+
"lstrip": false,
|
| 1479 |
+
"normalized": false,
|
| 1480 |
+
"rstrip": false,
|
| 1481 |
+
"single_word": false,
|
| 1482 |
+
"special": true
|
| 1483 |
+
},
|
| 1484 |
+
"128185": {
|
| 1485 |
+
"content": "<|reserved_special_token_177|>",
|
| 1486 |
+
"lstrip": false,
|
| 1487 |
+
"normalized": false,
|
| 1488 |
+
"rstrip": false,
|
| 1489 |
+
"single_word": false,
|
| 1490 |
+
"special": true
|
| 1491 |
+
},
|
| 1492 |
+
"128186": {
|
| 1493 |
+
"content": "<|reserved_special_token_178|>",
|
| 1494 |
+
"lstrip": false,
|
| 1495 |
+
"normalized": false,
|
| 1496 |
+
"rstrip": false,
|
| 1497 |
+
"single_word": false,
|
| 1498 |
+
"special": true
|
| 1499 |
+
},
|
| 1500 |
+
"128187": {
|
| 1501 |
+
"content": "<|reserved_special_token_179|>",
|
| 1502 |
+
"lstrip": false,
|
| 1503 |
+
"normalized": false,
|
| 1504 |
+
"rstrip": false,
|
| 1505 |
+
"single_word": false,
|
| 1506 |
+
"special": true
|
| 1507 |
+
},
|
| 1508 |
+
"128188": {
|
| 1509 |
+
"content": "<|reserved_special_token_180|>",
|
| 1510 |
+
"lstrip": false,
|
| 1511 |
+
"normalized": false,
|
| 1512 |
+
"rstrip": false,
|
| 1513 |
+
"single_word": false,
|
| 1514 |
+
"special": true
|
| 1515 |
+
},
|
| 1516 |
+
"128189": {
|
| 1517 |
+
"content": "<|reserved_special_token_181|>",
|
| 1518 |
+
"lstrip": false,
|
| 1519 |
+
"normalized": false,
|
| 1520 |
+
"rstrip": false,
|
| 1521 |
+
"single_word": false,
|
| 1522 |
+
"special": true
|
| 1523 |
+
},
|
| 1524 |
+
"128190": {
|
| 1525 |
+
"content": "<|reserved_special_token_182|>",
|
| 1526 |
+
"lstrip": false,
|
| 1527 |
+
"normalized": false,
|
| 1528 |
+
"rstrip": false,
|
| 1529 |
+
"single_word": false,
|
| 1530 |
+
"special": true
|
| 1531 |
+
},
|
| 1532 |
+
"128191": {
|
| 1533 |
+
"content": "<|reserved_special_token_183|>",
|
| 1534 |
+
"lstrip": false,
|
| 1535 |
+
"normalized": false,
|
| 1536 |
+
"rstrip": false,
|
| 1537 |
+
"single_word": false,
|
| 1538 |
+
"special": true
|
| 1539 |
+
},
|
| 1540 |
+
"128192": {
|
| 1541 |
+
"content": "<|reserved_special_token_184|>",
|
| 1542 |
+
"lstrip": false,
|
| 1543 |
+
"normalized": false,
|
| 1544 |
+
"rstrip": false,
|
| 1545 |
+
"single_word": false,
|
| 1546 |
+
"special": true
|
| 1547 |
+
},
|
| 1548 |
+
"128193": {
|
| 1549 |
+
"content": "<|reserved_special_token_185|>",
|
| 1550 |
+
"lstrip": false,
|
| 1551 |
+
"normalized": false,
|
| 1552 |
+
"rstrip": false,
|
| 1553 |
+
"single_word": false,
|
| 1554 |
+
"special": true
|
| 1555 |
+
},
|
| 1556 |
+
"128194": {
|
| 1557 |
+
"content": "<|reserved_special_token_186|>",
|
| 1558 |
+
"lstrip": false,
|
| 1559 |
+
"normalized": false,
|
| 1560 |
+
"rstrip": false,
|
| 1561 |
+
"single_word": false,
|
| 1562 |
+
"special": true
|
| 1563 |
+
},
|
| 1564 |
+
"128195": {
|
| 1565 |
+
"content": "<|reserved_special_token_187|>",
|
| 1566 |
+
"lstrip": false,
|
| 1567 |
+
"normalized": false,
|
| 1568 |
+
"rstrip": false,
|
| 1569 |
+
"single_word": false,
|
| 1570 |
+
"special": true
|
| 1571 |
+
},
|
| 1572 |
+
"128196": {
|
| 1573 |
+
"content": "<|reserved_special_token_188|>",
|
| 1574 |
+
"lstrip": false,
|
| 1575 |
+
"normalized": false,
|
| 1576 |
+
"rstrip": false,
|
| 1577 |
+
"single_word": false,
|
| 1578 |
+
"special": true
|
| 1579 |
+
},
|
| 1580 |
+
"128197": {
|
| 1581 |
+
"content": "<|reserved_special_token_189|>",
|
| 1582 |
+
"lstrip": false,
|
| 1583 |
+
"normalized": false,
|
| 1584 |
+
"rstrip": false,
|
| 1585 |
+
"single_word": false,
|
| 1586 |
+
"special": true
|
| 1587 |
+
},
|
| 1588 |
+
"128198": {
|
| 1589 |
+
"content": "<|reserved_special_token_190|>",
|
| 1590 |
+
"lstrip": false,
|
| 1591 |
+
"normalized": false,
|
| 1592 |
+
"rstrip": false,
|
| 1593 |
+
"single_word": false,
|
| 1594 |
+
"special": true
|
| 1595 |
+
},
|
| 1596 |
+
"128199": {
|
| 1597 |
+
"content": "<|reserved_special_token_191|>",
|
| 1598 |
+
"lstrip": false,
|
| 1599 |
+
"normalized": false,
|
| 1600 |
+
"rstrip": false,
|
| 1601 |
+
"single_word": false,
|
| 1602 |
+
"special": true
|
| 1603 |
+
},
|
| 1604 |
+
"128200": {
|
| 1605 |
+
"content": "<|reserved_special_token_192|>",
|
| 1606 |
+
"lstrip": false,
|
| 1607 |
+
"normalized": false,
|
| 1608 |
+
"rstrip": false,
|
| 1609 |
+
"single_word": false,
|
| 1610 |
+
"special": true
|
| 1611 |
+
},
|
| 1612 |
+
"128201": {
|
| 1613 |
+
"content": "<|reserved_special_token_193|>",
|
| 1614 |
+
"lstrip": false,
|
| 1615 |
+
"normalized": false,
|
| 1616 |
+
"rstrip": false,
|
| 1617 |
+
"single_word": false,
|
| 1618 |
+
"special": true
|
| 1619 |
+
},
|
| 1620 |
+
"128202": {
|
| 1621 |
+
"content": "<|reserved_special_token_194|>",
|
| 1622 |
+
"lstrip": false,
|
| 1623 |
+
"normalized": false,
|
| 1624 |
+
"rstrip": false,
|
| 1625 |
+
"single_word": false,
|
| 1626 |
+
"special": true
|
| 1627 |
+
},
|
| 1628 |
+
"128203": {
|
| 1629 |
+
"content": "<|reserved_special_token_195|>",
|
| 1630 |
+
"lstrip": false,
|
| 1631 |
+
"normalized": false,
|
| 1632 |
+
"rstrip": false,
|
| 1633 |
+
"single_word": false,
|
| 1634 |
+
"special": true
|
| 1635 |
+
},
|
| 1636 |
+
"128204": {
|
| 1637 |
+
"content": "<|reserved_special_token_196|>",
|
| 1638 |
+
"lstrip": false,
|
| 1639 |
+
"normalized": false,
|
| 1640 |
+
"rstrip": false,
|
| 1641 |
+
"single_word": false,
|
| 1642 |
+
"special": true
|
| 1643 |
+
},
|
| 1644 |
+
"128205": {
|
| 1645 |
+
"content": "<|reserved_special_token_197|>",
|
| 1646 |
+
"lstrip": false,
|
| 1647 |
+
"normalized": false,
|
| 1648 |
+
"rstrip": false,
|
| 1649 |
+
"single_word": false,
|
| 1650 |
+
"special": true
|
| 1651 |
+
},
|
| 1652 |
+
"128206": {
|
| 1653 |
+
"content": "<|reserved_special_token_198|>",
|
| 1654 |
+
"lstrip": false,
|
| 1655 |
+
"normalized": false,
|
| 1656 |
+
"rstrip": false,
|
| 1657 |
+
"single_word": false,
|
| 1658 |
+
"special": true
|
| 1659 |
+
},
|
| 1660 |
+
"128207": {
|
| 1661 |
+
"content": "<|reserved_special_token_199|>",
|
| 1662 |
+
"lstrip": false,
|
| 1663 |
+
"normalized": false,
|
| 1664 |
+
"rstrip": false,
|
| 1665 |
+
"single_word": false,
|
| 1666 |
+
"special": true
|
| 1667 |
+
},
|
| 1668 |
+
"128208": {
|
| 1669 |
+
"content": "<|reserved_special_token_200|>",
|
| 1670 |
+
"lstrip": false,
|
| 1671 |
+
"normalized": false,
|
| 1672 |
+
"rstrip": false,
|
| 1673 |
+
"single_word": false,
|
| 1674 |
+
"special": true
|
| 1675 |
+
},
|
| 1676 |
+
"128209": {
|
| 1677 |
+
"content": "<|reserved_special_token_201|>",
|
| 1678 |
+
"lstrip": false,
|
| 1679 |
+
"normalized": false,
|
| 1680 |
+
"rstrip": false,
|
| 1681 |
+
"single_word": false,
|
| 1682 |
+
"special": true
|
| 1683 |
+
},
|
| 1684 |
+
"128210": {
|
| 1685 |
+
"content": "<|reserved_special_token_202|>",
|
| 1686 |
+
"lstrip": false,
|
| 1687 |
+
"normalized": false,
|
| 1688 |
+
"rstrip": false,
|
| 1689 |
+
"single_word": false,
|
| 1690 |
+
"special": true
|
| 1691 |
+
},
|
| 1692 |
+
"128211": {
|
| 1693 |
+
"content": "<|reserved_special_token_203|>",
|
| 1694 |
+
"lstrip": false,
|
| 1695 |
+
"normalized": false,
|
| 1696 |
+
"rstrip": false,
|
| 1697 |
+
"single_word": false,
|
| 1698 |
+
"special": true
|
| 1699 |
+
},
|
| 1700 |
+
"128212": {
|
| 1701 |
+
"content": "<|reserved_special_token_204|>",
|
| 1702 |
+
"lstrip": false,
|
| 1703 |
+
"normalized": false,
|
| 1704 |
+
"rstrip": false,
|
| 1705 |
+
"single_word": false,
|
| 1706 |
+
"special": true
|
| 1707 |
+
},
|
| 1708 |
+
"128213": {
|
| 1709 |
+
"content": "<|reserved_special_token_205|>",
|
| 1710 |
+
"lstrip": false,
|
| 1711 |
+
"normalized": false,
|
| 1712 |
+
"rstrip": false,
|
| 1713 |
+
"single_word": false,
|
| 1714 |
+
"special": true
|
| 1715 |
+
},
|
| 1716 |
+
"128214": {
|
| 1717 |
+
"content": "<|reserved_special_token_206|>",
|
| 1718 |
+
"lstrip": false,
|
| 1719 |
+
"normalized": false,
|
| 1720 |
+
"rstrip": false,
|
| 1721 |
+
"single_word": false,
|
| 1722 |
+
"special": true
|
| 1723 |
+
},
|
| 1724 |
+
"128215": {
|
| 1725 |
+
"content": "<|reserved_special_token_207|>",
|
| 1726 |
+
"lstrip": false,
|
| 1727 |
+
"normalized": false,
|
| 1728 |
+
"rstrip": false,
|
| 1729 |
+
"single_word": false,
|
| 1730 |
+
"special": true
|
| 1731 |
+
},
|
| 1732 |
+
"128216": {
|
| 1733 |
+
"content": "<|reserved_special_token_208|>",
|
| 1734 |
+
"lstrip": false,
|
| 1735 |
+
"normalized": false,
|
| 1736 |
+
"rstrip": false,
|
| 1737 |
+
"single_word": false,
|
| 1738 |
+
"special": true
|
| 1739 |
+
},
|
| 1740 |
+
"128217": {
|
| 1741 |
+
"content": "<|reserved_special_token_209|>",
|
| 1742 |
+
"lstrip": false,
|
| 1743 |
+
"normalized": false,
|
| 1744 |
+
"rstrip": false,
|
| 1745 |
+
"single_word": false,
|
| 1746 |
+
"special": true
|
| 1747 |
+
},
|
| 1748 |
+
"128218": {
|
| 1749 |
+
"content": "<|reserved_special_token_210|>",
|
| 1750 |
+
"lstrip": false,
|
| 1751 |
+
"normalized": false,
|
| 1752 |
+
"rstrip": false,
|
| 1753 |
+
"single_word": false,
|
| 1754 |
+
"special": true
|
| 1755 |
+
},
|
| 1756 |
+
"128219": {
|
| 1757 |
+
"content": "<|reserved_special_token_211|>",
|
| 1758 |
+
"lstrip": false,
|
| 1759 |
+
"normalized": false,
|
| 1760 |
+
"rstrip": false,
|
| 1761 |
+
"single_word": false,
|
| 1762 |
+
"special": true
|
| 1763 |
+
},
|
| 1764 |
+
"128220": {
|
| 1765 |
+
"content": "<|reserved_special_token_212|>",
|
| 1766 |
+
"lstrip": false,
|
| 1767 |
+
"normalized": false,
|
| 1768 |
+
"rstrip": false,
|
| 1769 |
+
"single_word": false,
|
| 1770 |
+
"special": true
|
| 1771 |
+
},
|
| 1772 |
+
"128221": {
|
| 1773 |
+
"content": "<|reserved_special_token_213|>",
|
| 1774 |
+
"lstrip": false,
|
| 1775 |
+
"normalized": false,
|
| 1776 |
+
"rstrip": false,
|
| 1777 |
+
"single_word": false,
|
| 1778 |
+
"special": true
|
| 1779 |
+
},
|
| 1780 |
+
"128222": {
|
| 1781 |
+
"content": "<|reserved_special_token_214|>",
|
| 1782 |
+
"lstrip": false,
|
| 1783 |
+
"normalized": false,
|
| 1784 |
+
"rstrip": false,
|
| 1785 |
+
"single_word": false,
|
| 1786 |
+
"special": true
|
| 1787 |
+
},
|
| 1788 |
+
"128223": {
|
| 1789 |
+
"content": "<|reserved_special_token_215|>",
|
| 1790 |
+
"lstrip": false,
|
| 1791 |
+
"normalized": false,
|
| 1792 |
+
"rstrip": false,
|
| 1793 |
+
"single_word": false,
|
| 1794 |
+
"special": true
|
| 1795 |
+
},
|
| 1796 |
+
"128224": {
|
| 1797 |
+
"content": "<|reserved_special_token_216|>",
|
| 1798 |
+
"lstrip": false,
|
| 1799 |
+
"normalized": false,
|
| 1800 |
+
"rstrip": false,
|
| 1801 |
+
"single_word": false,
|
| 1802 |
+
"special": true
|
| 1803 |
+
},
|
| 1804 |
+
"128225": {
|
| 1805 |
+
"content": "<|reserved_special_token_217|>",
|
| 1806 |
+
"lstrip": false,
|
| 1807 |
+
"normalized": false,
|
| 1808 |
+
"rstrip": false,
|
| 1809 |
+
"single_word": false,
|
| 1810 |
+
"special": true
|
| 1811 |
+
},
|
| 1812 |
+
"128226": {
|
| 1813 |
+
"content": "<|reserved_special_token_218|>",
|
| 1814 |
+
"lstrip": false,
|
| 1815 |
+
"normalized": false,
|
| 1816 |
+
"rstrip": false,
|
| 1817 |
+
"single_word": false,
|
| 1818 |
+
"special": true
|
| 1819 |
+
},
|
| 1820 |
+
"128227": {
|
| 1821 |
+
"content": "<|reserved_special_token_219|>",
|
| 1822 |
+
"lstrip": false,
|
| 1823 |
+
"normalized": false,
|
| 1824 |
+
"rstrip": false,
|
| 1825 |
+
"single_word": false,
|
| 1826 |
+
"special": true
|
| 1827 |
+
},
|
| 1828 |
+
"128228": {
|
| 1829 |
+
"content": "<|reserved_special_token_220|>",
|
| 1830 |
+
"lstrip": false,
|
| 1831 |
+
"normalized": false,
|
| 1832 |
+
"rstrip": false,
|
| 1833 |
+
"single_word": false,
|
| 1834 |
+
"special": true
|
| 1835 |
+
},
|
| 1836 |
+
"128229": {
|
| 1837 |
+
"content": "<|reserved_special_token_221|>",
|
| 1838 |
+
"lstrip": false,
|
| 1839 |
+
"normalized": false,
|
| 1840 |
+
"rstrip": false,
|
| 1841 |
+
"single_word": false,
|
| 1842 |
+
"special": true
|
| 1843 |
+
},
|
| 1844 |
+
"128230": {
|
| 1845 |
+
"content": "<|reserved_special_token_222|>",
|
| 1846 |
+
"lstrip": false,
|
| 1847 |
+
"normalized": false,
|
| 1848 |
+
"rstrip": false,
|
| 1849 |
+
"single_word": false,
|
| 1850 |
+
"special": true
|
| 1851 |
+
},
|
| 1852 |
+
"128231": {
|
| 1853 |
+
"content": "<|reserved_special_token_223|>",
|
| 1854 |
+
"lstrip": false,
|
| 1855 |
+
"normalized": false,
|
| 1856 |
+
"rstrip": false,
|
| 1857 |
+
"single_word": false,
|
| 1858 |
+
"special": true
|
| 1859 |
+
},
|
| 1860 |
+
"128232": {
|
| 1861 |
+
"content": "<|reserved_special_token_224|>",
|
| 1862 |
+
"lstrip": false,
|
| 1863 |
+
"normalized": false,
|
| 1864 |
+
"rstrip": false,
|
| 1865 |
+
"single_word": false,
|
| 1866 |
+
"special": true
|
| 1867 |
+
},
|
| 1868 |
+
"128233": {
|
| 1869 |
+
"content": "<|reserved_special_token_225|>",
|
| 1870 |
+
"lstrip": false,
|
| 1871 |
+
"normalized": false,
|
| 1872 |
+
"rstrip": false,
|
| 1873 |
+
"single_word": false,
|
| 1874 |
+
"special": true
|
| 1875 |
+
},
|
| 1876 |
+
"128234": {
|
| 1877 |
+
"content": "<|reserved_special_token_226|>",
|
| 1878 |
+
"lstrip": false,
|
| 1879 |
+
"normalized": false,
|
| 1880 |
+
"rstrip": false,
|
| 1881 |
+
"single_word": false,
|
| 1882 |
+
"special": true
|
| 1883 |
+
},
|
| 1884 |
+
"128235": {
|
| 1885 |
+
"content": "<|reserved_special_token_227|>",
|
| 1886 |
+
"lstrip": false,
|
| 1887 |
+
"normalized": false,
|
| 1888 |
+
"rstrip": false,
|
| 1889 |
+
"single_word": false,
|
| 1890 |
+
"special": true
|
| 1891 |
+
},
|
| 1892 |
+
"128236": {
|
| 1893 |
+
"content": "<|reserved_special_token_228|>",
|
| 1894 |
+
"lstrip": false,
|
| 1895 |
+
"normalized": false,
|
| 1896 |
+
"rstrip": false,
|
| 1897 |
+
"single_word": false,
|
| 1898 |
+
"special": true
|
| 1899 |
+
},
|
| 1900 |
+
"128237": {
|
| 1901 |
+
"content": "<|reserved_special_token_229|>",
|
| 1902 |
+
"lstrip": false,
|
| 1903 |
+
"normalized": false,
|
| 1904 |
+
"rstrip": false,
|
| 1905 |
+
"single_word": false,
|
| 1906 |
+
"special": true
|
| 1907 |
+
},
|
| 1908 |
+
"128238": {
|
| 1909 |
+
"content": "<|reserved_special_token_230|>",
|
| 1910 |
+
"lstrip": false,
|
| 1911 |
+
"normalized": false,
|
| 1912 |
+
"rstrip": false,
|
| 1913 |
+
"single_word": false,
|
| 1914 |
+
"special": true
|
| 1915 |
+
},
|
| 1916 |
+
"128239": {
|
| 1917 |
+
"content": "<|reserved_special_token_231|>",
|
| 1918 |
+
"lstrip": false,
|
| 1919 |
+
"normalized": false,
|
| 1920 |
+
"rstrip": false,
|
| 1921 |
+
"single_word": false,
|
| 1922 |
+
"special": true
|
| 1923 |
+
},
|
| 1924 |
+
"128240": {
|
| 1925 |
+
"content": "<|reserved_special_token_232|>",
|
| 1926 |
+
"lstrip": false,
|
| 1927 |
+
"normalized": false,
|
| 1928 |
+
"rstrip": false,
|
| 1929 |
+
"single_word": false,
|
| 1930 |
+
"special": true
|
| 1931 |
+
},
|
| 1932 |
+
"128241": {
|
| 1933 |
+
"content": "<|reserved_special_token_233|>",
|
| 1934 |
+
"lstrip": false,
|
| 1935 |
+
"normalized": false,
|
| 1936 |
+
"rstrip": false,
|
| 1937 |
+
"single_word": false,
|
| 1938 |
+
"special": true
|
| 1939 |
+
},
|
| 1940 |
+
"128242": {
|
| 1941 |
+
"content": "<|reserved_special_token_234|>",
|
| 1942 |
+
"lstrip": false,
|
| 1943 |
+
"normalized": false,
|
| 1944 |
+
"rstrip": false,
|
| 1945 |
+
"single_word": false,
|
| 1946 |
+
"special": true
|
| 1947 |
+
},
|
| 1948 |
+
"128243": {
|
| 1949 |
+
"content": "<|reserved_special_token_235|>",
|
| 1950 |
+
"lstrip": false,
|
| 1951 |
+
"normalized": false,
|
| 1952 |
+
"rstrip": false,
|
| 1953 |
+
"single_word": false,
|
| 1954 |
+
"special": true
|
| 1955 |
+
},
|
| 1956 |
+
"128244": {
|
| 1957 |
+
"content": "<|reserved_special_token_236|>",
|
| 1958 |
+
"lstrip": false,
|
| 1959 |
+
"normalized": false,
|
| 1960 |
+
"rstrip": false,
|
| 1961 |
+
"single_word": false,
|
| 1962 |
+
"special": true
|
| 1963 |
+
},
|
| 1964 |
+
"128245": {
|
| 1965 |
+
"content": "<|reserved_special_token_237|>",
|
| 1966 |
+
"lstrip": false,
|
| 1967 |
+
"normalized": false,
|
| 1968 |
+
"rstrip": false,
|
| 1969 |
+
"single_word": false,
|
| 1970 |
+
"special": true
|
| 1971 |
+
},
|
| 1972 |
+
"128246": {
|
| 1973 |
+
"content": "<|reserved_special_token_238|>",
|
| 1974 |
+
"lstrip": false,
|
| 1975 |
+
"normalized": false,
|
| 1976 |
+
"rstrip": false,
|
| 1977 |
+
"single_word": false,
|
| 1978 |
+
"special": true
|
| 1979 |
+
},
|
| 1980 |
+
"128247": {
|
| 1981 |
+
"content": "<|reserved_special_token_239|>",
|
| 1982 |
+
"lstrip": false,
|
| 1983 |
+
"normalized": false,
|
| 1984 |
+
"rstrip": false,
|
| 1985 |
+
"single_word": false,
|
| 1986 |
+
"special": true
|
| 1987 |
+
},
|
| 1988 |
+
"128248": {
|
| 1989 |
+
"content": "<|reserved_special_token_240|>",
|
| 1990 |
+
"lstrip": false,
|
| 1991 |
+
"normalized": false,
|
| 1992 |
+
"rstrip": false,
|
| 1993 |
+
"single_word": false,
|
| 1994 |
+
"special": true
|
| 1995 |
+
},
|
| 1996 |
+
"128249": {
|
| 1997 |
+
"content": "<|reserved_special_token_241|>",
|
| 1998 |
+
"lstrip": false,
|
| 1999 |
+
"normalized": false,
|
| 2000 |
+
"rstrip": false,
|
| 2001 |
+
"single_word": false,
|
| 2002 |
+
"special": true
|
| 2003 |
+
},
|
| 2004 |
+
"128250": {
|
| 2005 |
+
"content": "<|reserved_special_token_242|>",
|
| 2006 |
+
"lstrip": false,
|
| 2007 |
+
"normalized": false,
|
| 2008 |
+
"rstrip": false,
|
| 2009 |
+
"single_word": false,
|
| 2010 |
+
"special": true
|
| 2011 |
+
},
|
| 2012 |
+
"128251": {
|
| 2013 |
+
"content": "<|reserved_special_token_243|>",
|
| 2014 |
+
"lstrip": false,
|
| 2015 |
+
"normalized": false,
|
| 2016 |
+
"rstrip": false,
|
| 2017 |
+
"single_word": false,
|
| 2018 |
+
"special": true
|
| 2019 |
+
},
|
| 2020 |
+
"128252": {
|
| 2021 |
+
"content": "<|reserved_special_token_244|>",
|
| 2022 |
+
"lstrip": false,
|
| 2023 |
+
"normalized": false,
|
| 2024 |
+
"rstrip": false,
|
| 2025 |
+
"single_word": false,
|
| 2026 |
+
"special": true
|
| 2027 |
+
},
|
| 2028 |
+
"128253": {
|
| 2029 |
+
"content": "<|reserved_special_token_245|>",
|
| 2030 |
+
"lstrip": false,
|
| 2031 |
+
"normalized": false,
|
| 2032 |
+
"rstrip": false,
|
| 2033 |
+
"single_word": false,
|
| 2034 |
+
"special": true
|
| 2035 |
+
},
|
| 2036 |
+
"128254": {
|
| 2037 |
+
"content": "<|reserved_special_token_246|>",
|
| 2038 |
+
"lstrip": false,
|
| 2039 |
+
"normalized": false,
|
| 2040 |
+
"rstrip": false,
|
| 2041 |
+
"single_word": false,
|
| 2042 |
+
"special": true
|
| 2043 |
+
},
|
| 2044 |
+
"128255": {
|
| 2045 |
+
"content": "<|reserved_special_token_247|>",
|
| 2046 |
+
"lstrip": false,
|
| 2047 |
+
"normalized": false,
|
| 2048 |
+
"rstrip": false,
|
| 2049 |
+
"single_word": false,
|
| 2050 |
+
"special": true
|
| 2051 |
+
}
|
| 2052 |
+
},
|
| 2053 |
+
"bos_token": "<|begin_of_text|>",
|
| 2054 |
+
"clean_up_tokenization_spaces": true,
|
| 2055 |
+
"eos_token": "<|eot_id|>",
|
| 2056 |
+
"extra_special_tokens": {},
|
| 2057 |
+
"max_length": null,
|
| 2058 |
+
"model_input_names": [
|
| 2059 |
+
"input_ids",
|
| 2060 |
+
"attention_mask"
|
| 2061 |
+
],
|
| 2062 |
+
"model_max_length": 131072,
|
| 2063 |
+
"pad_to_multiple_of": null,
|
| 2064 |
+
"pad_token": "<|finetune_right_pad_id|>",
|
| 2065 |
+
"pad_token_type_id": 0,
|
| 2066 |
+
"padding_side": "right",
|
| 2067 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 2068 |
+
"unk_token": null
|
| 2069 |
+
}
|
v3-results/heldout_cases.json
ADDED
|
@@ -0,0 +1,392 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"id": "heldout_00",
|
| 4 |
+
"prompt": "Return only the integer: 701 + 91 - 31.",
|
| 5 |
+
"expected": "761",
|
| 6 |
+
"kind": "integer"
|
| 7 |
+
},
|
| 8 |
+
{
|
| 9 |
+
"id": "heldout_01",
|
| 10 |
+
"prompt": "A crate has 704 red and 92 blue beads. Remove 32 red beads. Return only the total beads remaining.",
|
| 11 |
+
"expected": "764",
|
| 12 |
+
"kind": "integer"
|
| 13 |
+
},
|
| 14 |
+
{
|
| 15 |
+
"id": "heldout_02",
|
| 16 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_707, count=93.",
|
| 17 |
+
"expected": {
|
| 18 |
+
"item": "part_707",
|
| 19 |
+
"count": 93
|
| 20 |
+
},
|
| 21 |
+
"kind": "json"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"id": "heldout_03",
|
| 25 |
+
"prompt": "Return only one CSV value row, no header: name=Item710; quantity=94; code=K34.",
|
| 26 |
+
"expected": "Item710,94,K34",
|
| 27 |
+
"kind": "exact"
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"id": "heldout_04",
|
| 31 |
+
"prompt": "Copy exactly, no other text: Key_713-B95_C35",
|
| 32 |
+
"expected": "Key_713-B95_C35",
|
| 33 |
+
"kind": "exact"
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"id": "heldout_05",
|
| 37 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 96 + 36.",
|
| 38 |
+
"expected": "x * 96 + 36",
|
| 39 |
+
"kind": "code"
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"id": "heldout_06",
|
| 43 |
+
"prompt": "Return only the integer: 719 + 97 - 37.",
|
| 44 |
+
"expected": "779",
|
| 45 |
+
"kind": "integer"
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"id": "heldout_07",
|
| 49 |
+
"prompt": "A crate has 722 red and 98 blue beads. Remove 38 red beads. Return only the total beads remaining.",
|
| 50 |
+
"expected": "782",
|
| 51 |
+
"kind": "integer"
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"id": "heldout_08",
|
| 55 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_725, count=99.",
|
| 56 |
+
"expected": {
|
| 57 |
+
"item": "part_725",
|
| 58 |
+
"count": 99
|
| 59 |
+
},
|
| 60 |
+
"kind": "json"
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"id": "heldout_09",
|
| 64 |
+
"prompt": "Return only one CSV value row, no header: name=Item728; quantity=100; code=K40.",
|
| 65 |
+
"expected": "Item728,100,K40",
|
| 66 |
+
"kind": "exact"
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"id": "heldout_10",
|
| 70 |
+
"prompt": "Copy exactly, no other text: Key_731-B101_C41",
|
| 71 |
+
"expected": "Key_731-B101_C41",
|
| 72 |
+
"kind": "exact"
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"id": "heldout_11",
|
| 76 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 102 + 42.",
|
| 77 |
+
"expected": "x * 102 + 42",
|
| 78 |
+
"kind": "code"
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"id": "heldout_12",
|
| 82 |
+
"prompt": "Return only the integer: 737 + 103 - 43.",
|
| 83 |
+
"expected": "797",
|
| 84 |
+
"kind": "integer"
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"id": "heldout_13",
|
| 88 |
+
"prompt": "A crate has 740 red and 104 blue beads. Remove 44 red beads. Return only the total beads remaining.",
|
| 89 |
+
"expected": "800",
|
| 90 |
+
"kind": "integer"
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"id": "heldout_14",
|
| 94 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_743, count=105.",
|
| 95 |
+
"expected": {
|
| 96 |
+
"item": "part_743",
|
| 97 |
+
"count": 105
|
| 98 |
+
},
|
| 99 |
+
"kind": "json"
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"id": "heldout_15",
|
| 103 |
+
"prompt": "Return only one CSV value row, no header: name=Item746; quantity=106; code=K46.",
|
| 104 |
+
"expected": "Item746,106,K46",
|
| 105 |
+
"kind": "exact"
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"id": "heldout_16",
|
| 109 |
+
"prompt": "Copy exactly, no other text: Key_749-B107_C47",
|
| 110 |
+
"expected": "Key_749-B107_C47",
|
| 111 |
+
"kind": "exact"
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"id": "heldout_17",
|
| 115 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 108 + 31.",
|
| 116 |
+
"expected": "x * 108 + 31",
|
| 117 |
+
"kind": "code"
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"id": "heldout_18",
|
| 121 |
+
"prompt": "Return only the integer: 755 + 109 - 32.",
|
| 122 |
+
"expected": "832",
|
| 123 |
+
"kind": "integer"
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"id": "heldout_19",
|
| 127 |
+
"prompt": "A crate has 758 red and 110 blue beads. Remove 33 red beads. Return only the total beads remaining.",
|
| 128 |
+
"expected": "835",
|
| 129 |
+
"kind": "integer"
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"id": "heldout_20",
|
| 133 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_761, count=111.",
|
| 134 |
+
"expected": {
|
| 135 |
+
"item": "part_761",
|
| 136 |
+
"count": 111
|
| 137 |
+
},
|
| 138 |
+
"kind": "json"
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"id": "heldout_21",
|
| 142 |
+
"prompt": "Return only one CSV value row, no header: name=Item764; quantity=112; code=K35.",
|
| 143 |
+
"expected": "Item764,112,K35",
|
| 144 |
+
"kind": "exact"
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"id": "heldout_22",
|
| 148 |
+
"prompt": "Copy exactly, no other text: Key_767-B113_C36",
|
| 149 |
+
"expected": "Key_767-B113_C36",
|
| 150 |
+
"kind": "exact"
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"id": "heldout_23",
|
| 154 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 114 + 37.",
|
| 155 |
+
"expected": "x * 114 + 37",
|
| 156 |
+
"kind": "code"
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"id": "heldout_24",
|
| 160 |
+
"prompt": "Return only the integer: 773 + 115 - 38.",
|
| 161 |
+
"expected": "850",
|
| 162 |
+
"kind": "integer"
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"id": "heldout_25",
|
| 166 |
+
"prompt": "A crate has 776 red and 116 blue beads. Remove 39 red beads. Return only the total beads remaining.",
|
| 167 |
+
"expected": "853",
|
| 168 |
+
"kind": "integer"
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"id": "heldout_26",
|
| 172 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_779, count=117.",
|
| 173 |
+
"expected": {
|
| 174 |
+
"item": "part_779",
|
| 175 |
+
"count": 117
|
| 176 |
+
},
|
| 177 |
+
"kind": "json"
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"id": "heldout_27",
|
| 181 |
+
"prompt": "Return only one CSV value row, no header: name=Item782; quantity=118; code=K41.",
|
| 182 |
+
"expected": "Item782,118,K41",
|
| 183 |
+
"kind": "exact"
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"id": "heldout_28",
|
| 187 |
+
"prompt": "Copy exactly, no other text: Key_785-B119_C42",
|
| 188 |
+
"expected": "Key_785-B119_C42",
|
| 189 |
+
"kind": "exact"
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"id": "heldout_29",
|
| 193 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 120 + 43.",
|
| 194 |
+
"expected": "x * 120 + 43",
|
| 195 |
+
"kind": "code"
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"id": "heldout_30",
|
| 199 |
+
"prompt": "Return only the integer: 791 + 121 - 44.",
|
| 200 |
+
"expected": "868",
|
| 201 |
+
"kind": "integer"
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"id": "heldout_31",
|
| 205 |
+
"prompt": "A crate has 794 red and 122 blue beads. Remove 45 red beads. Return only the total beads remaining.",
|
| 206 |
+
"expected": "871",
|
| 207 |
+
"kind": "integer"
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"id": "heldout_32",
|
| 211 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_797, count=123.",
|
| 212 |
+
"expected": {
|
| 213 |
+
"item": "part_797",
|
| 214 |
+
"count": 123
|
| 215 |
+
},
|
| 216 |
+
"kind": "json"
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"id": "heldout_33",
|
| 220 |
+
"prompt": "Return only one CSV value row, no header: name=Item800; quantity=124; code=K47.",
|
| 221 |
+
"expected": "Item800,124,K47",
|
| 222 |
+
"kind": "exact"
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"id": "heldout_34",
|
| 226 |
+
"prompt": "Copy exactly, no other text: Key_803-B125_C31",
|
| 227 |
+
"expected": "Key_803-B125_C31",
|
| 228 |
+
"kind": "exact"
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"id": "heldout_35",
|
| 232 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 126 + 32.",
|
| 233 |
+
"expected": "x * 126 + 32",
|
| 234 |
+
"kind": "code"
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"id": "heldout_36",
|
| 238 |
+
"prompt": "Return only the integer: 809 + 127 - 33.",
|
| 239 |
+
"expected": "903",
|
| 240 |
+
"kind": "integer"
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"id": "heldout_37",
|
| 244 |
+
"prompt": "A crate has 812 red and 128 blue beads. Remove 34 red beads. Return only the total beads remaining.",
|
| 245 |
+
"expected": "906",
|
| 246 |
+
"kind": "integer"
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"id": "heldout_38",
|
| 250 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_815, count=129.",
|
| 251 |
+
"expected": {
|
| 252 |
+
"item": "part_815",
|
| 253 |
+
"count": 129
|
| 254 |
+
},
|
| 255 |
+
"kind": "json"
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"id": "heldout_39",
|
| 259 |
+
"prompt": "Return only one CSV value row, no header: name=Item818; quantity=130; code=K36.",
|
| 260 |
+
"expected": "Item818,130,K36",
|
| 261 |
+
"kind": "exact"
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"id": "heldout_40",
|
| 265 |
+
"prompt": "Copy exactly, no other text: Key_821-B131_C37",
|
| 266 |
+
"expected": "Key_821-B131_C37",
|
| 267 |
+
"kind": "exact"
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"id": "heldout_41",
|
| 271 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 132 + 38.",
|
| 272 |
+
"expected": "x * 132 + 38",
|
| 273 |
+
"kind": "code"
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"id": "heldout_42",
|
| 277 |
+
"prompt": "Return only the integer: 827 + 133 - 39.",
|
| 278 |
+
"expected": "921",
|
| 279 |
+
"kind": "integer"
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"id": "heldout_43",
|
| 283 |
+
"prompt": "A crate has 830 red and 134 blue beads. Remove 40 red beads. Return only the total beads remaining.",
|
| 284 |
+
"expected": "924",
|
| 285 |
+
"kind": "integer"
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"id": "heldout_44",
|
| 289 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_833, count=135.",
|
| 290 |
+
"expected": {
|
| 291 |
+
"item": "part_833",
|
| 292 |
+
"count": 135
|
| 293 |
+
},
|
| 294 |
+
"kind": "json"
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"id": "heldout_45",
|
| 298 |
+
"prompt": "Return only one CSV value row, no header: name=Item836; quantity=136; code=K42.",
|
| 299 |
+
"expected": "Item836,136,K42",
|
| 300 |
+
"kind": "exact"
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"id": "heldout_46",
|
| 304 |
+
"prompt": "Copy exactly, no other text: Key_839-B137_C43",
|
| 305 |
+
"expected": "Key_839-B137_C43",
|
| 306 |
+
"kind": "exact"
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"id": "heldout_47",
|
| 310 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 138 + 44.",
|
| 311 |
+
"expected": "x * 138 + 44",
|
| 312 |
+
"kind": "code"
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"id": "heldout_48",
|
| 316 |
+
"prompt": "Return only the integer: 845 + 139 - 45.",
|
| 317 |
+
"expected": "939",
|
| 318 |
+
"kind": "integer"
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"id": "heldout_49",
|
| 322 |
+
"prompt": "A crate has 848 red and 140 blue beads. Remove 46 red beads. Return only the total beads remaining.",
|
| 323 |
+
"expected": "942",
|
| 324 |
+
"kind": "integer"
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"id": "heldout_50",
|
| 328 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_851, count=141.",
|
| 329 |
+
"expected": {
|
| 330 |
+
"item": "part_851",
|
| 331 |
+
"count": 141
|
| 332 |
+
},
|
| 333 |
+
"kind": "json"
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"id": "heldout_51",
|
| 337 |
+
"prompt": "Return only one CSV value row, no header: name=Item854; quantity=142; code=K31.",
|
| 338 |
+
"expected": "Item854,142,K31",
|
| 339 |
+
"kind": "exact"
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"id": "heldout_52",
|
| 343 |
+
"prompt": "Copy exactly, no other text: Key_857-B143_C32",
|
| 344 |
+
"expected": "Key_857-B143_C32",
|
| 345 |
+
"kind": "exact"
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"id": "heldout_53",
|
| 349 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 144 + 33.",
|
| 350 |
+
"expected": "x * 144 + 33",
|
| 351 |
+
"kind": "code"
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"id": "heldout_54",
|
| 355 |
+
"prompt": "Return only the integer: 863 + 145 - 34.",
|
| 356 |
+
"expected": "974",
|
| 357 |
+
"kind": "integer"
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"id": "heldout_55",
|
| 361 |
+
"prompt": "A crate has 866 red and 146 blue beads. Remove 35 red beads. Return only the total beads remaining.",
|
| 362 |
+
"expected": "977",
|
| 363 |
+
"kind": "integer"
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"id": "heldout_56",
|
| 367 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_869, count=147.",
|
| 368 |
+
"expected": {
|
| 369 |
+
"item": "part_869",
|
| 370 |
+
"count": 147
|
| 371 |
+
},
|
| 372 |
+
"kind": "json"
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"id": "heldout_57",
|
| 376 |
+
"prompt": "Return only one CSV value row, no header: name=Item872; quantity=148; code=K37.",
|
| 377 |
+
"expected": "Item872,148,K37",
|
| 378 |
+
"kind": "exact"
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"id": "heldout_58",
|
| 382 |
+
"prompt": "Copy exactly, no other text: Key_875-B149_C38",
|
| 383 |
+
"expected": "Key_875-B149_C38",
|
| 384 |
+
"kind": "exact"
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"id": "heldout_59",
|
| 388 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 150 + 39.",
|
| 389 |
+
"expected": "x * 150 + 39",
|
| 390 |
+
"kind": "code"
|
| 391 |
+
}
|
| 392 |
+
]
|
v3-results/train_v3.py
ADDED
|
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# /// script
|
| 2 |
+
# dependencies = ["torch==2.8.0", "transformers==4.56.2", "peft==0.17.1", "accelerate==1.10.1", "huggingface_hub==0.34.4", "bitsandbytes==0.47.0", "safetensors==0.6.2"]
|
| 3 |
+
# ///
|
| 4 |
+
"""Budget-bounded V3 training with paired, disjoint evaluation and reload gate."""
|
| 5 |
+
import os, json, random, time, re, gc, hashlib, traceback
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
import torch
|
| 8 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, Trainer, TrainingArguments, TrainerCallback, set_seed
|
| 9 |
+
from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training, PeftModel
|
| 10 |
+
from huggingface_hub import HfApi, hf_hub_download
|
| 11 |
+
|
| 12 |
+
REPO='SEE7/Astheral'; REV='d9aa899722be4788f57107c2951c72d0ec6614ee'
|
| 13 |
+
OUT=Path('/tmp/astheral-v3'); OUT.mkdir(exist_ok=True)
|
| 14 |
+
api=HfApi(); started=time.monotonic(); set_seed(20261113); rng=random.Random(20261113)
|
| 15 |
+
assert torch.cuda.is_available() and 'A10G' in torch.cuda.get_device_name(0)
|
| 16 |
+
branch='v3-candidate-20261113'
|
| 17 |
+
api.create_branch(REPO,branch=branch,revision=REV,exist_ok=True)
|
| 18 |
+
settings=json.loads(Path(hf_hub_download(REPO,'settings.json',revision=REV)).read_text())
|
| 19 |
+
tok=AutoTokenizer.from_pretrained(REPO,revision=REV); tok.pad_token_id=128004
|
| 20 |
+
report={'status':'preparing','parent_revision':REV,'hardware':torch.cuda.get_device_name(0),'job_timeout_seconds':5400,'listed_usd_per_hour':1.0,'training_seed':20261113}
|
| 21 |
+
|
| 22 |
+
def publish():
|
| 23 |
+
report['elapsed_seconds']=round(time.monotonic()-started,2)
|
| 24 |
+
(OUT/'v3_report.json').write_text(json.dumps(report,indent=2))
|
| 25 |
+
api.upload_folder(repo_id=REPO,folder_path=OUT,revision=branch,path_in_repo='v3-results',ignore_patterns=['merged/**'],commit_message='Record measured V3 training and evaluation artifacts')
|
| 26 |
+
|
| 27 |
+
def sample(a,b,c,kind):
|
| 28 |
+
if kind==0:return f'Return only the integer: {a} + {b} - {c}.',str(a+b-c),'integer'
|
| 29 |
+
if kind==1:return f'A crate has {a} red and {b} blue beads. Remove {c} red beads. Return only the total beads remaining.',str(a+b-c),'integer'
|
| 30 |
+
if kind==2:return f'Return only valid JSON with keys "item" and "count": item=part_{a}, count={b}.',{'item':f'part_{a}','count':b},'json'
|
| 31 |
+
if kind==3:return f'Return only one CSV value row, no header: name=Item{a}; quantity={b}; code=K{c}.',f'Item{a},{b},K{c}','exact'
|
| 32 |
+
if kind==4:return f'Copy exactly, no other text: Key_{a}-B{b}_C{c}',f'Key_{a}-B{b}_C{c}','exact'
|
| 33 |
+
return f'Write only a Python function named solve(x) that returns x * {b} + {c}.',f'x * {b} + {c}','code'
|
| 34 |
+
|
| 35 |
+
def grade(case,text):
|
| 36 |
+
expected=case['expected']; kind=case['kind']; text=text.strip()
|
| 37 |
+
try:
|
| 38 |
+
if kind=='json':return json.loads(text)==expected
|
| 39 |
+
if kind=='integer':return re.fullmatch(r'-?\d+',text) is not None and int(text)==int(expected)
|
| 40 |
+
if kind=='exact':return text==expected
|
| 41 |
+
# Parse and evaluate a tiny arithmetic-only AST; never execute model-generated code.
|
| 42 |
+
import ast
|
| 43 |
+
clean=re.sub(r'^```(?:python)?\s*|\s*```$','',text).strip();tree=ast.parse(clean)
|
| 44 |
+
if len(tree.body)!=1 or not isinstance(tree.body[0],ast.FunctionDef):return False
|
| 45 |
+
fn=tree.body[0]
|
| 46 |
+
if fn.name!='solve' or len(fn.args.args)!=1 or fn.args.args[0].arg!='x' or len(fn.body)!=1 or not isinstance(fn.body[0],ast.Return):return False
|
| 47 |
+
def calc(n,x):
|
| 48 |
+
if isinstance(n,ast.Constant) and isinstance(n.value,(int,float)):return n.value
|
| 49 |
+
if isinstance(n,ast.Name) and n.id=='x':return x
|
| 50 |
+
if isinstance(n,ast.BinOp) and isinstance(n.op,(ast.Add,ast.Sub,ast.Mult)):
|
| 51 |
+
a,b=calc(n.left,x),calc(n.right,x)
|
| 52 |
+
return a+b if isinstance(n.op,ast.Add) else a-b if isinstance(n.op,ast.Sub) else a*b
|
| 53 |
+
raise ValueError('Unsupported expression')
|
| 54 |
+
match=re.fullmatch(r'x \* (\d+) \+ (\d+)',expected);b,c=map(int,match.groups())
|
| 55 |
+
return all(calc(fn.body[0].value,x)==x*b+c for x in [-5,0,1,7,100])
|
| 56 |
+
except Exception:return False
|
| 57 |
+
|
| 58 |
+
tests=[]
|
| 59 |
+
for i in range(60):
|
| 60 |
+
# Evaluation numbers are outside the synthetic training ranges.
|
| 61 |
+
a,b,c=701+i*3,91+i,31+i%17;p,e,k=sample(a,b,c,i%6)
|
| 62 |
+
tests.append({'id':f'heldout_{i:02}','prompt':p,'expected':e,'kind':k})
|
| 63 |
+
train=[]
|
| 64 |
+
for i in range(900):
|
| 65 |
+
a,b,c=rng.randint(100,500),rng.randint(10,80),rng.randint(1,29)
|
| 66 |
+
p,e,k=sample(a,b,c,i%6)
|
| 67 |
+
answer=json.dumps(e) if k=='json' else 'def solve(x):\n return '+e if k=='code' else str(e)
|
| 68 |
+
train.append({'messages':[{'role':'user','content':p},{'role':'assistant','content':answer}]})
|
| 69 |
+
retained=[json.loads(s) for s in Path(hf_hub_download(REPO,'train.jsonl',revision=REV)).read_text().splitlines()]
|
| 70 |
+
rng.shuffle(retained);train+=retained[:900];rng.shuffle(train)
|
| 71 |
+
assert not {x['prompt'] for x in tests} & {m['content'] for row in train for m in row['messages'] if m['role']=='user'}
|
| 72 |
+
(OUT/'heldout_cases.json').write_text(json.dumps(tests,indent=2));(OUT/'training.jsonl').write_text('\n'.join(json.dumps(x) for x in train)+'\n')
|
| 73 |
+
(OUT/'train_v3.py').write_text(Path(__file__).read_text())
|
| 74 |
+
q=BitsAndBytesConfig(load_in_4bit=True,bnb_4bit_quant_type='nf4',bnb_4bit_use_double_quant=True,bnb_4bit_compute_dtype=torch.bfloat16)
|
| 75 |
+
model=AutoModelForCausalLM.from_pretrained(REPO,revision=REV,quantization_config=q,device_map={'':'cuda:0'},torch_dtype=torch.bfloat16,attn_implementation='sdpa')
|
| 76 |
+
|
| 77 |
+
@torch.inference_mode()
|
| 78 |
+
def evaluate(label):
|
| 79 |
+
model.eval(); rows=[]
|
| 80 |
+
for i,case in enumerate(tests):
|
| 81 |
+
inputs=tok.apply_chat_template([{'role':'system','content':settings['system']},{'role':'user','content':case['prompt']}],tokenize=True,add_generation_prompt=True,return_dict=True,return_tensors='pt').to('cuda')
|
| 82 |
+
out=model.generate(**inputs,max_new_tokens=96,do_sample=False,pad_token_id=128004,eos_token_id=128009)
|
| 83 |
+
text=tok.decode(out[0,inputs['input_ids'].shape[1]:],skip_special_tokens=True)
|
| 84 |
+
rows.append({**case,'output':text,'pass':grade(case,text),'tokens':int(out.shape[1]-inputs['input_ids'].shape[1])})
|
| 85 |
+
print('EVAL',label,i,rows[-1]['pass'],flush=True)
|
| 86 |
+
(OUT/(label+'_outputs.json')).write_text(json.dumps(rows,indent=2))
|
| 87 |
+
return rows
|
| 88 |
+
|
| 89 |
+
try:
|
| 90 |
+
baseline=evaluate('v2_quantized'); report['baseline_passed']=sum(r['pass'] for r in baseline);publish()
|
| 91 |
+
model=prepare_model_for_kbit_training(model,use_gradient_checkpointing=True)
|
| 92 |
+
model=get_peft_model(model,LoraConfig(r=16,lora_alpha=32,lora_dropout=0.05,target_modules=['q_proj','k_proj','v_proj','o_proj','gate_proj','up_proj','down_proj'],task_type='CAUSAL_LM'))
|
| 93 |
+
model.config.use_cache=False;encoded=[]
|
| 94 |
+
for row in train:
|
| 95 |
+
messages=row['messages']; prefix=tok.apply_chat_template(messages[:-1],tokenize=True,add_generation_prompt=True)
|
| 96 |
+
full=tok.apply_chat_template(messages,tokenize=True,add_generation_prompt=False)
|
| 97 |
+
if len(full)<=512 and full[:len(prefix)]==prefix:encoded.append({'input_ids':full,'labels':[-100]*len(prefix)+full[len(prefix):]})
|
| 98 |
+
def collate(rows):
|
| 99 |
+
length=max(len(r['input_ids']) for r in rows)
|
| 100 |
+
return {k:torch.tensor([r[k]+[pad]*(length-len(r[k])) for r in rows]) for k,pad in [('input_ids',128004),('labels',-100)]}|{'attention_mask':torch.tensor([[1]*len(r['input_ids'])+[0]*(length-len(r['input_ids'])) for r in rows])}
|
| 101 |
+
class Budget(TrainerCallback):
|
| 102 |
+
def on_step_end(self,args,state,control,**kwargs):
|
| 103 |
+
if time.monotonic()-started>2400:control.should_training_stop=True
|
| 104 |
+
return control
|
| 105 |
+
args=TrainingArguments(output_dir='/tmp/v3-checkpoints',max_steps=180,per_device_train_batch_size=1,gradient_accumulation_steps=4,learning_rate=1e-5,warmup_ratio=0.05,bf16=True,gradient_checkpointing=True,gradient_checkpointing_kwargs={'use_reentrant':False},optim='paged_adamw_8bit',save_strategy='no',logging_steps=10,report_to='none',remove_unused_columns=False,seed=20261113)
|
| 106 |
+
trainer=Trainer(model=model,args=args,train_dataset=encoded,data_collator=collate,callbacks=[Budget()])
|
| 107 |
+
report.update(status='training',training_examples=len(encoded),trainable_parameters=sum(p.numel() for p in model.parameters() if p.requires_grad));publish()
|
| 108 |
+
result=trainer.train();report['training']=result.metrics;report['optimizer_steps']=trainer.state.global_step
|
| 109 |
+
adapter=OUT/'adapter';model.save_pretrained(adapter,safe_serialization=True);tok.save_pretrained(adapter)
|
| 110 |
+
model.config.use_cache=True;model.gradient_checkpointing_disable();candidate=evaluate('v3_quantized')
|
| 111 |
+
report['candidate_passed']=sum(r['pass'] for r in candidate);report['status']='trained';publish()
|
| 112 |
+
del trainer,model;gc.collect();torch.cuda.empty_cache()
|
| 113 |
+
model=AutoModelForCausalLM.from_pretrained(REPO,revision=REV,torch_dtype=torch.bfloat16,device_map={'':'cuda:0'},attn_implementation='sdpa')
|
| 114 |
+
model=PeftModel.from_pretrained(model,adapter).merge_and_unload(safe_merge=True)
|
| 115 |
+
model.config.astheral_version='3';merged=OUT/'merged';model.save_pretrained(merged,safe_serialization=True,max_shard_size='4GB');tok.save_pretrained(merged)
|
| 116 |
+
del model;gc.collect();torch.cuda.empty_cache()
|
| 117 |
+
model=AutoModelForCausalLM.from_pretrained(merged,torch_dtype=torch.bfloat16,device_map={'':'cuda:0'},attn_implementation='sdpa')
|
| 118 |
+
reloaded=evaluate('v3_reloaded');report['reloaded_passed']=sum(r['pass'] for r in reloaded)
|
| 119 |
+
groups={kind:{'baseline':sum(r['pass'] for r in baseline if r['kind']==kind),'candidate':sum(r['pass'] for r in reloaded if r['kind']==kind),'total':sum(r['kind']==kind for r in tests)} for kind in {x['kind'] for x in tests}}
|
| 120 |
+
report['groups']=groups;report['quality_gate_passed']=report['reloaded_passed']>report['baseline_passed'] and all(v['candidate']>=v['baseline'] for v in groups.values())
|
| 121 |
+
report['status']='complete';report['peak_gpu_bytes']=torch.cuda.max_memory_allocated();publish()
|
| 122 |
+
# Save the actual trained weights to the isolated candidate branch even on a failed gate.
|
| 123 |
+
api.upload_folder(repo_id=REPO,folder_path=merged,revision=branch,commit_message='Save reloaded V3 candidate weights; see measured quality gate in v3-results')
|
| 124 |
+
if report['quality_gate_passed']:
|
| 125 |
+
api.create_tag(REPO,tag='v3',revision=api.model_info(REPO,revision=branch).sha,exist_ok=True)
|
| 126 |
+
print('V3_COMPLETE',json.dumps(report),flush=True)
|
| 127 |
+
except Exception:
|
| 128 |
+
report['status']='failed';report['error']=traceback.format_exc();publish();raise
|
v3-results/training.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
v3-results/v2_quantized_outputs.json
ADDED
|
@@ -0,0 +1,572 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"id": "heldout_00",
|
| 4 |
+
"prompt": "Return only the integer: 701 + 91 - 31.",
|
| 5 |
+
"expected": "761",
|
| 6 |
+
"kind": "integer",
|
| 7 |
+
"output": "701 + 91 - 31 is 761.",
|
| 8 |
+
"pass": false,
|
| 9 |
+
"tokens": 12
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"id": "heldout_01",
|
| 13 |
+
"prompt": "A crate has 704 red and 92 blue beads. Remove 32 red beads. Return only the total beads remaining.",
|
| 14 |
+
"expected": "764",
|
| 15 |
+
"kind": "integer",
|
| 16 |
+
"output": "Remove 32 from 704 to find 672 red beads. 672 + 92 = 764 beads remain.",
|
| 17 |
+
"pass": false,
|
| 18 |
+
"tokens": 25
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"id": "heldout_02",
|
| 22 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_707, count=93.",
|
| 23 |
+
"expected": {
|
| 24 |
+
"item": "part_707",
|
| 25 |
+
"count": 93
|
| 26 |
+
},
|
| 27 |
+
"kind": "json",
|
| 28 |
+
"output": "{\"item\": \"part_707\", \"count\": 93}",
|
| 29 |
+
"pass": true,
|
| 30 |
+
"tokens": 15
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"id": "heldout_03",
|
| 34 |
+
"prompt": "Return only one CSV value row, no header: name=Item710; quantity=94; code=K34.",
|
| 35 |
+
"expected": "Item710,94,K34",
|
| 36 |
+
"kind": "exact",
|
| 37 |
+
"output": "Item710,94,K34",
|
| 38 |
+
"pass": true,
|
| 39 |
+
"tokens": 7
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"id": "heldout_04",
|
| 43 |
+
"prompt": "Copy exactly, no other text: Key_713-B95_C35",
|
| 44 |
+
"expected": "Key_713-B95_C35",
|
| 45 |
+
"kind": "exact",
|
| 46 |
+
"output": "Key_713-B95_C35",
|
| 47 |
+
"pass": true,
|
| 48 |
+
"tokens": 8
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"id": "heldout_05",
|
| 52 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 96 + 36.",
|
| 53 |
+
"expected": "x * 96 + 36",
|
| 54 |
+
"kind": "code",
|
| 55 |
+
"output": "```python\ndef solve(x):\n return x * 96 + 36\n```",
|
| 56 |
+
"pass": true,
|
| 57 |
+
"tokens": 19
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"id": "heldout_06",
|
| 61 |
+
"prompt": "Return only the integer: 719 + 97 - 37.",
|
| 62 |
+
"expected": "779",
|
| 63 |
+
"kind": "integer",
|
| 64 |
+
"output": "719 + 97 - 37 = 779",
|
| 65 |
+
"pass": false,
|
| 66 |
+
"tokens": 11
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"id": "heldout_07",
|
| 70 |
+
"prompt": "A crate has 722 red and 98 blue beads. Remove 38 red beads. Return only the total beads remaining.",
|
| 71 |
+
"expected": "782",
|
| 72 |
+
"kind": "integer",
|
| 73 |
+
"output": "Remove 38 red beads from 722, leaving 684. The 98 blue beads are unchanged, for a total of 684 + 98 = 782 beads.",
|
| 74 |
+
"pass": false,
|
| 75 |
+
"tokens": 36
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"id": "heldout_08",
|
| 79 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_725, count=99.",
|
| 80 |
+
"expected": {
|
| 81 |
+
"item": "part_725",
|
| 82 |
+
"count": 99
|
| 83 |
+
},
|
| 84 |
+
"kind": "json",
|
| 85 |
+
"output": "{\"item\": \"part_725\", \"count\": 99}",
|
| 86 |
+
"pass": true,
|
| 87 |
+
"tokens": 15
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"id": "heldout_09",
|
| 91 |
+
"prompt": "Return only one CSV value row, no header: name=Item728; quantity=100; code=K40.",
|
| 92 |
+
"expected": "Item728,100,K40",
|
| 93 |
+
"kind": "exact",
|
| 94 |
+
"output": "Item728,100,K40",
|
| 95 |
+
"pass": true,
|
| 96 |
+
"tokens": 7
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"id": "heldout_10",
|
| 100 |
+
"prompt": "Copy exactly, no other text: Key_731-B101_C41",
|
| 101 |
+
"expected": "Key_731-B101_C41",
|
| 102 |
+
"kind": "exact",
|
| 103 |
+
"output": "Key_731-B101_C41",
|
| 104 |
+
"pass": true,
|
| 105 |
+
"tokens": 8
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"id": "heldout_11",
|
| 109 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 102 + 42.",
|
| 110 |
+
"expected": "x * 102 + 42",
|
| 111 |
+
"kind": "code",
|
| 112 |
+
"output": "```python\ndef solve(x):\n return x * 102 + 42\n```",
|
| 113 |
+
"pass": true,
|
| 114 |
+
"tokens": 19
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"id": "heldout_12",
|
| 118 |
+
"prompt": "Return only the integer: 737 + 103 - 43.",
|
| 119 |
+
"expected": "797",
|
| 120 |
+
"kind": "integer",
|
| 121 |
+
"output": "737 + 103 - 43 is 797.",
|
| 122 |
+
"pass": false,
|
| 123 |
+
"tokens": 12
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"id": "heldout_13",
|
| 127 |
+
"prompt": "A crate has 740 red and 104 blue beads. Remove 44 red beads. Return only the total beads remaining.",
|
| 128 |
+
"expected": "800",
|
| 129 |
+
"kind": "integer",
|
| 130 |
+
"output": "Remove 44 red beads from 740, leaving 696. The 104 blue beads are unchanged, for 696 + 104 = 800 beads remaining.",
|
| 131 |
+
"pass": false,
|
| 132 |
+
"tokens": 34
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"id": "heldout_14",
|
| 136 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_743, count=105.",
|
| 137 |
+
"expected": {
|
| 138 |
+
"item": "part_743",
|
| 139 |
+
"count": 105
|
| 140 |
+
},
|
| 141 |
+
"kind": "json",
|
| 142 |
+
"output": "{\"item\": \"part_743\", \"count\": 105}",
|
| 143 |
+
"pass": true,
|
| 144 |
+
"tokens": 15
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"id": "heldout_15",
|
| 148 |
+
"prompt": "Return only one CSV value row, no header: name=Item746; quantity=106; code=K46.",
|
| 149 |
+
"expected": "Item746,106,K46",
|
| 150 |
+
"kind": "exact",
|
| 151 |
+
"output": "Item746,106,K46",
|
| 152 |
+
"pass": true,
|
| 153 |
+
"tokens": 7
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"id": "heldout_16",
|
| 157 |
+
"prompt": "Copy exactly, no other text: Key_749-B107_C47",
|
| 158 |
+
"expected": "Key_749-B107_C47",
|
| 159 |
+
"kind": "exact",
|
| 160 |
+
"output": "Key_749-B107_C47",
|
| 161 |
+
"pass": true,
|
| 162 |
+
"tokens": 8
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"id": "heldout_17",
|
| 166 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 108 + 31.",
|
| 167 |
+
"expected": "x * 108 + 31",
|
| 168 |
+
"kind": "code",
|
| 169 |
+
"output": "```python\ndef solve(x):\n return x * 108 + 31\n```",
|
| 170 |
+
"pass": true,
|
| 171 |
+
"tokens": 19
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"id": "heldout_18",
|
| 175 |
+
"prompt": "Return only the integer: 755 + 109 - 32.",
|
| 176 |
+
"expected": "832",
|
| 177 |
+
"kind": "integer",
|
| 178 |
+
"output": "755 + 109 - 32 is 832.",
|
| 179 |
+
"pass": false,
|
| 180 |
+
"tokens": 12
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"id": "heldout_19",
|
| 184 |
+
"prompt": "A crate has 758 red and 110 blue beads. Remove 33 red beads. Return only the total beads remaining.",
|
| 185 |
+
"expected": "835",
|
| 186 |
+
"kind": "integer",
|
| 187 |
+
"output": "Remove 33 red beads from 758, leaving 725. The 110 blue beads are unchanged, for 725 + 110 = 835 beads remaining.",
|
| 188 |
+
"pass": false,
|
| 189 |
+
"tokens": 34
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"id": "heldout_20",
|
| 193 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_761, count=111.",
|
| 194 |
+
"expected": {
|
| 195 |
+
"item": "part_761",
|
| 196 |
+
"count": 111
|
| 197 |
+
},
|
| 198 |
+
"kind": "json",
|
| 199 |
+
"output": "{\"item\": \"part_761\", \"count\": 111}",
|
| 200 |
+
"pass": true,
|
| 201 |
+
"tokens": 15
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"id": "heldout_21",
|
| 205 |
+
"prompt": "Return only one CSV value row, no header: name=Item764; quantity=112; code=K35.",
|
| 206 |
+
"expected": "Item764,112,K35",
|
| 207 |
+
"kind": "exact",
|
| 208 |
+
"output": "Item764,112,K35",
|
| 209 |
+
"pass": true,
|
| 210 |
+
"tokens": 7
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"id": "heldout_22",
|
| 214 |
+
"prompt": "Copy exactly, no other text: Key_767-B113_C36",
|
| 215 |
+
"expected": "Key_767-B113_C36",
|
| 216 |
+
"kind": "exact",
|
| 217 |
+
"output": "Key_767-B113_C36",
|
| 218 |
+
"pass": true,
|
| 219 |
+
"tokens": 8
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"id": "heldout_23",
|
| 223 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 114 + 37.",
|
| 224 |
+
"expected": "x * 114 + 37",
|
| 225 |
+
"kind": "code",
|
| 226 |
+
"output": "```python\ndef solve(x):\n return x * 114 + 37\n```",
|
| 227 |
+
"pass": true,
|
| 228 |
+
"tokens": 19
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"id": "heldout_24",
|
| 232 |
+
"prompt": "Return only the integer: 773 + 115 - 38.",
|
| 233 |
+
"expected": "850",
|
| 234 |
+
"kind": "integer",
|
| 235 |
+
"output": "750",
|
| 236 |
+
"pass": false,
|
| 237 |
+
"tokens": 2
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"id": "heldout_25",
|
| 241 |
+
"prompt": "A crate has 776 red and 116 blue beads. Remove 39 red beads. Return only the total beads remaining.",
|
| 242 |
+
"expected": "853",
|
| 243 |
+
"kind": "integer",
|
| 244 |
+
"output": "Remove 39 red beads from 776 to leave 737. The 116 blue beads are unchanged, for 737 + 116 = 853 beads remaining.",
|
| 245 |
+
"pass": false,
|
| 246 |
+
"tokens": 34
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"id": "heldout_26",
|
| 250 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_779, count=117.",
|
| 251 |
+
"expected": {
|
| 252 |
+
"item": "part_779",
|
| 253 |
+
"count": 117
|
| 254 |
+
},
|
| 255 |
+
"kind": "json",
|
| 256 |
+
"output": "{\"item\": \"part_779\", \"count\": 117}",
|
| 257 |
+
"pass": true,
|
| 258 |
+
"tokens": 15
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"id": "heldout_27",
|
| 262 |
+
"prompt": "Return only one CSV value row, no header: name=Item782; quantity=118; code=K41.",
|
| 263 |
+
"expected": "Item782,118,K41",
|
| 264 |
+
"kind": "exact",
|
| 265 |
+
"output": "Item782,118,K41",
|
| 266 |
+
"pass": true,
|
| 267 |
+
"tokens": 7
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"id": "heldout_28",
|
| 271 |
+
"prompt": "Copy exactly, no other text: Key_785-B119_C42",
|
| 272 |
+
"expected": "Key_785-B119_C42",
|
| 273 |
+
"kind": "exact",
|
| 274 |
+
"output": "Key_785-B119_C42",
|
| 275 |
+
"pass": true,
|
| 276 |
+
"tokens": 8
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"id": "heldout_29",
|
| 280 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 120 + 43.",
|
| 281 |
+
"expected": "x * 120 + 43",
|
| 282 |
+
"kind": "code",
|
| 283 |
+
"output": "```python\ndef solve(x):\n return x * 120 + 43\n```",
|
| 284 |
+
"pass": true,
|
| 285 |
+
"tokens": 19
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"id": "heldout_30",
|
| 289 |
+
"prompt": "Return only the integer: 791 + 121 - 44.",
|
| 290 |
+
"expected": "868",
|
| 291 |
+
"kind": "integer",
|
| 292 |
+
"output": "791 + 121 - 44 is 868.",
|
| 293 |
+
"pass": false,
|
| 294 |
+
"tokens": 12
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"id": "heldout_31",
|
| 298 |
+
"prompt": "A crate has 794 red and 122 blue beads. Remove 45 red beads. Return only the total beads remaining.",
|
| 299 |
+
"expected": "871",
|
| 300 |
+
"kind": "integer",
|
| 301 |
+
"output": "Remove 45 red beads from 794, leaving 749, and keep the 122 blue beads, leaving 749 + 122 = 871 beads.",
|
| 302 |
+
"pass": false,
|
| 303 |
+
"tokens": 33
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"id": "heldout_32",
|
| 307 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_797, count=123.",
|
| 308 |
+
"expected": {
|
| 309 |
+
"item": "part_797",
|
| 310 |
+
"count": 123
|
| 311 |
+
},
|
| 312 |
+
"kind": "json",
|
| 313 |
+
"output": "{\"item\":\"part_797\",\"count\":123}",
|
| 314 |
+
"pass": true,
|
| 315 |
+
"tokens": 12
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"id": "heldout_33",
|
| 319 |
+
"prompt": "Return only one CSV value row, no header: name=Item800; quantity=124; code=K47.",
|
| 320 |
+
"expected": "Item800,124,K47",
|
| 321 |
+
"kind": "exact",
|
| 322 |
+
"output": "Item800,124,K47",
|
| 323 |
+
"pass": true,
|
| 324 |
+
"tokens": 7
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"id": "heldout_34",
|
| 328 |
+
"prompt": "Copy exactly, no other text: Key_803-B125_C31",
|
| 329 |
+
"expected": "Key_803-B125_C31",
|
| 330 |
+
"kind": "exact",
|
| 331 |
+
"output": "Key_803-B125_C31",
|
| 332 |
+
"pass": true,
|
| 333 |
+
"tokens": 8
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"id": "heldout_35",
|
| 337 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 126 + 32.",
|
| 338 |
+
"expected": "x * 126 + 32",
|
| 339 |
+
"kind": "code",
|
| 340 |
+
"output": "```python\ndef solve(x):\n return x * 126 + 32\n```",
|
| 341 |
+
"pass": true,
|
| 342 |
+
"tokens": 19
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"id": "heldout_36",
|
| 346 |
+
"prompt": "Return only the integer: 809 + 127 - 33.",
|
| 347 |
+
"expected": "903",
|
| 348 |
+
"kind": "integer",
|
| 349 |
+
"output": "809 + 127 - 33 is 903.",
|
| 350 |
+
"pass": false,
|
| 351 |
+
"tokens": 12
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"id": "heldout_37",
|
| 355 |
+
"prompt": "A crate has 812 red and 128 blue beads. Remove 34 red beads. Return only the total beads remaining.",
|
| 356 |
+
"expected": "906",
|
| 357 |
+
"kind": "integer",
|
| 358 |
+
"output": "Remove 34 red beads from 812, leaving 778, and keep 128 blue beads, for 778 + 128 = 906 beads.",
|
| 359 |
+
"pass": false,
|
| 360 |
+
"tokens": 32
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"id": "heldout_38",
|
| 364 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_815, count=129.",
|
| 365 |
+
"expected": {
|
| 366 |
+
"item": "part_815",
|
| 367 |
+
"count": 129
|
| 368 |
+
},
|
| 369 |
+
"kind": "json",
|
| 370 |
+
"output": "{\"item\":\"part_815\",\"count\":129}",
|
| 371 |
+
"pass": true,
|
| 372 |
+
"tokens": 12
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"id": "heldout_39",
|
| 376 |
+
"prompt": "Return only one CSV value row, no header: name=Item818; quantity=130; code=K36.",
|
| 377 |
+
"expected": "Item818,130,K36",
|
| 378 |
+
"kind": "exact",
|
| 379 |
+
"output": "Item818,130,K36",
|
| 380 |
+
"pass": true,
|
| 381 |
+
"tokens": 7
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"id": "heldout_40",
|
| 385 |
+
"prompt": "Copy exactly, no other text: Key_821-B131_C37",
|
| 386 |
+
"expected": "Key_821-B131_C37",
|
| 387 |
+
"kind": "exact",
|
| 388 |
+
"output": "Key_821-B131_C37",
|
| 389 |
+
"pass": true,
|
| 390 |
+
"tokens": 8
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"id": "heldout_41",
|
| 394 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 132 + 38.",
|
| 395 |
+
"expected": "x * 132 + 38",
|
| 396 |
+
"kind": "code",
|
| 397 |
+
"output": "```python\ndef solve(x):\n return x * 132 + 38\n```",
|
| 398 |
+
"pass": true,
|
| 399 |
+
"tokens": 19
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"id": "heldout_42",
|
| 403 |
+
"prompt": "Return only the integer: 827 + 133 - 39.",
|
| 404 |
+
"expected": "921",
|
| 405 |
+
"kind": "integer",
|
| 406 |
+
"output": "827 + 133 - 39 = 921",
|
| 407 |
+
"pass": false,
|
| 408 |
+
"tokens": 11
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"id": "heldout_43",
|
| 412 |
+
"prompt": "A crate has 830 red and 134 blue beads. Remove 40 red beads. Return only the total beads remaining.",
|
| 413 |
+
"expected": "924",
|
| 414 |
+
"kind": "integer",
|
| 415 |
+
"output": "Remove 40 red beads from 830, leaving 790. The 134 blue beads are unchanged, for 790 + 134 = 924 beads remaining.",
|
| 416 |
+
"pass": false,
|
| 417 |
+
"tokens": 34
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"id": "heldout_44",
|
| 421 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_833, count=135.",
|
| 422 |
+
"expected": {
|
| 423 |
+
"item": "part_833",
|
| 424 |
+
"count": 135
|
| 425 |
+
},
|
| 426 |
+
"kind": "json",
|
| 427 |
+
"output": "{\"item\": \"part_833\", \"count\": 135}",
|
| 428 |
+
"pass": true,
|
| 429 |
+
"tokens": 15
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"id": "heldout_45",
|
| 433 |
+
"prompt": "Return only one CSV value row, no header: name=Item836; quantity=136; code=K42.",
|
| 434 |
+
"expected": "Item836,136,K42",
|
| 435 |
+
"kind": "exact",
|
| 436 |
+
"output": "Item836,136,K42",
|
| 437 |
+
"pass": true,
|
| 438 |
+
"tokens": 7
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"id": "heldout_46",
|
| 442 |
+
"prompt": "Copy exactly, no other text: Key_839-B137_C43",
|
| 443 |
+
"expected": "Key_839-B137_C43",
|
| 444 |
+
"kind": "exact",
|
| 445 |
+
"output": "Key_839-B137_C43",
|
| 446 |
+
"pass": true,
|
| 447 |
+
"tokens": 8
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"id": "heldout_47",
|
| 451 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 138 + 44.",
|
| 452 |
+
"expected": "x * 138 + 44",
|
| 453 |
+
"kind": "code",
|
| 454 |
+
"output": "```python\ndef solve(x):\n return x * 138 + 44\n```",
|
| 455 |
+
"pass": true,
|
| 456 |
+
"tokens": 19
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"id": "heldout_48",
|
| 460 |
+
"prompt": "Return only the integer: 845 + 139 - 45.",
|
| 461 |
+
"expected": "939",
|
| 462 |
+
"kind": "integer",
|
| 463 |
+
"output": "845 + 139 - 45 = 939",
|
| 464 |
+
"pass": false,
|
| 465 |
+
"tokens": 11
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"id": "heldout_49",
|
| 469 |
+
"prompt": "A crate has 848 red and 140 blue beads. Remove 46 red beads. Return only the total beads remaining.",
|
| 470 |
+
"expected": "942",
|
| 471 |
+
"kind": "integer",
|
| 472 |
+
"output": "Remove 46 red beads from 848, leaving 802. The 140 blue beads are unchanged, for 802 + 140 = 942 beads remaining.",
|
| 473 |
+
"pass": false,
|
| 474 |
+
"tokens": 34
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"id": "heldout_50",
|
| 478 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_851, count=141.",
|
| 479 |
+
"expected": {
|
| 480 |
+
"item": "part_851",
|
| 481 |
+
"count": 141
|
| 482 |
+
},
|
| 483 |
+
"kind": "json",
|
| 484 |
+
"output": "{\"item\": \"part_851\", \"count\": 141}",
|
| 485 |
+
"pass": true,
|
| 486 |
+
"tokens": 15
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"id": "heldout_51",
|
| 490 |
+
"prompt": "Return only one CSV value row, no header: name=Item854; quantity=142; code=K31.",
|
| 491 |
+
"expected": "Item854,142,K31",
|
| 492 |
+
"kind": "exact",
|
| 493 |
+
"output": "Item854,142,K31",
|
| 494 |
+
"pass": true,
|
| 495 |
+
"tokens": 7
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"id": "heldout_52",
|
| 499 |
+
"prompt": "Copy exactly, no other text: Key_857-B143_C32",
|
| 500 |
+
"expected": "Key_857-B143_C32",
|
| 501 |
+
"kind": "exact",
|
| 502 |
+
"output": "Key_857-B143_C32",
|
| 503 |
+
"pass": true,
|
| 504 |
+
"tokens": 8
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"id": "heldout_53",
|
| 508 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 144 + 33.",
|
| 509 |
+
"expected": "x * 144 + 33",
|
| 510 |
+
"kind": "code",
|
| 511 |
+
"output": "```python\ndef solve(x):\n return x * 144 + 33\n```",
|
| 512 |
+
"pass": true,
|
| 513 |
+
"tokens": 19
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"id": "heldout_54",
|
| 517 |
+
"prompt": "Return only the integer: 863 + 145 - 34.",
|
| 518 |
+
"expected": "974",
|
| 519 |
+
"kind": "integer",
|
| 520 |
+
"output": "863 + 145 - 34 is 974.",
|
| 521 |
+
"pass": false,
|
| 522 |
+
"tokens": 12
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"id": "heldout_55",
|
| 526 |
+
"prompt": "A crate has 866 red and 146 blue beads. Remove 35 red beads. Return only the total beads remaining.",
|
| 527 |
+
"expected": "977",
|
| 528 |
+
"kind": "integer",
|
| 529 |
+
"output": "Remove 35 red beads from 866 to leave 831. The 146 blue beads are unchanged, for 831 + 146 = 977 beads remaining.",
|
| 530 |
+
"pass": false,
|
| 531 |
+
"tokens": 34
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"id": "heldout_56",
|
| 535 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_869, count=147.",
|
| 536 |
+
"expected": {
|
| 537 |
+
"item": "part_869",
|
| 538 |
+
"count": 147
|
| 539 |
+
},
|
| 540 |
+
"kind": "json",
|
| 541 |
+
"output": "{\"item\": \"part_869\", \"count\": 147}",
|
| 542 |
+
"pass": true,
|
| 543 |
+
"tokens": 15
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"id": "heldout_57",
|
| 547 |
+
"prompt": "Return only one CSV value row, no header: name=Item872; quantity=148; code=K37.",
|
| 548 |
+
"expected": "Item872,148,K37",
|
| 549 |
+
"kind": "exact",
|
| 550 |
+
"output": "Item872,148,K37",
|
| 551 |
+
"pass": true,
|
| 552 |
+
"tokens": 7
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"id": "heldout_58",
|
| 556 |
+
"prompt": "Copy exactly, no other text: Key_875-B149_C38",
|
| 557 |
+
"expected": "Key_875-B149_C38",
|
| 558 |
+
"kind": "exact",
|
| 559 |
+
"output": "Key_875-B149_C38",
|
| 560 |
+
"pass": true,
|
| 561 |
+
"tokens": 8
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"id": "heldout_59",
|
| 565 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 150 + 39.",
|
| 566 |
+
"expected": "x * 150 + 39",
|
| 567 |
+
"kind": "code",
|
| 568 |
+
"output": "```python\ndef solve(x):\n return x * 150 + 39\n```",
|
| 569 |
+
"pass": true,
|
| 570 |
+
"tokens": 19
|
| 571 |
+
}
|
| 572 |
+
]
|
v3-results/v3_quantized_outputs.json
ADDED
|
@@ -0,0 +1,572 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"id": "heldout_00",
|
| 4 |
+
"prompt": "Return only the integer: 701 + 91 - 31.",
|
| 5 |
+
"expected": "761",
|
| 6 |
+
"kind": "integer",
|
| 7 |
+
"output": "761",
|
| 8 |
+
"pass": true,
|
| 9 |
+
"tokens": 2
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"id": "heldout_01",
|
| 13 |
+
"prompt": "A crate has 704 red and 92 blue beads. Remove 32 red beads. Return only the total beads remaining.",
|
| 14 |
+
"expected": "764",
|
| 15 |
+
"kind": "integer",
|
| 16 |
+
"output": "676",
|
| 17 |
+
"pass": false,
|
| 18 |
+
"tokens": 2
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"id": "heldout_02",
|
| 22 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_707, count=93.",
|
| 23 |
+
"expected": {
|
| 24 |
+
"item": "part_707",
|
| 25 |
+
"count": 93
|
| 26 |
+
},
|
| 27 |
+
"kind": "json",
|
| 28 |
+
"output": "{\"item\": \"part_707\", \"count\": 93}",
|
| 29 |
+
"pass": true,
|
| 30 |
+
"tokens": 15
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"id": "heldout_03",
|
| 34 |
+
"prompt": "Return only one CSV value row, no header: name=Item710; quantity=94; code=K34.",
|
| 35 |
+
"expected": "Item710,94,K34",
|
| 36 |
+
"kind": "exact",
|
| 37 |
+
"output": "Item710,94,K34",
|
| 38 |
+
"pass": true,
|
| 39 |
+
"tokens": 7
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"id": "heldout_04",
|
| 43 |
+
"prompt": "Copy exactly, no other text: Key_713-B95_C35",
|
| 44 |
+
"expected": "Key_713-B95_C35",
|
| 45 |
+
"kind": "exact",
|
| 46 |
+
"output": "Key_713-B95_C35",
|
| 47 |
+
"pass": true,
|
| 48 |
+
"tokens": 8
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"id": "heldout_05",
|
| 52 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 96 + 36.",
|
| 53 |
+
"expected": "x * 96 + 36",
|
| 54 |
+
"kind": "code",
|
| 55 |
+
"output": "```python\ndef solve(x):\n return x * 96 + 36\n```",
|
| 56 |
+
"pass": true,
|
| 57 |
+
"tokens": 19
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"id": "heldout_06",
|
| 61 |
+
"prompt": "Return only the integer: 719 + 97 - 37.",
|
| 62 |
+
"expected": "779",
|
| 63 |
+
"kind": "integer",
|
| 64 |
+
"output": "779",
|
| 65 |
+
"pass": true,
|
| 66 |
+
"tokens": 2
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"id": "heldout_07",
|
| 70 |
+
"prompt": "A crate has 722 red and 98 blue beads. Remove 38 red beads. Return only the total beads remaining.",
|
| 71 |
+
"expected": "782",
|
| 72 |
+
"kind": "integer",
|
| 73 |
+
"output": "722 - 38 = 684 red beads remain. 684 + 98 = 782 beads remain.",
|
| 74 |
+
"pass": false,
|
| 75 |
+
"tokens": 23
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"id": "heldout_08",
|
| 79 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_725, count=99.",
|
| 80 |
+
"expected": {
|
| 81 |
+
"item": "part_725",
|
| 82 |
+
"count": 99
|
| 83 |
+
},
|
| 84 |
+
"kind": "json",
|
| 85 |
+
"output": "{\"item\": \"part_725\", \"count\": 99}",
|
| 86 |
+
"pass": true,
|
| 87 |
+
"tokens": 15
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"id": "heldout_09",
|
| 91 |
+
"prompt": "Return only one CSV value row, no header: name=Item728; quantity=100; code=K40.",
|
| 92 |
+
"expected": "Item728,100,K40",
|
| 93 |
+
"kind": "exact",
|
| 94 |
+
"output": "Item728,100,K40",
|
| 95 |
+
"pass": true,
|
| 96 |
+
"tokens": 7
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"id": "heldout_10",
|
| 100 |
+
"prompt": "Copy exactly, no other text: Key_731-B101_C41",
|
| 101 |
+
"expected": "Key_731-B101_C41",
|
| 102 |
+
"kind": "exact",
|
| 103 |
+
"output": "Key_731-B101_C41",
|
| 104 |
+
"pass": true,
|
| 105 |
+
"tokens": 8
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"id": "heldout_11",
|
| 109 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 102 + 42.",
|
| 110 |
+
"expected": "x * 102 + 42",
|
| 111 |
+
"kind": "code",
|
| 112 |
+
"output": "```python\ndef solve(x):\n return x * 102 + 42\n```",
|
| 113 |
+
"pass": true,
|
| 114 |
+
"tokens": 19
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"id": "heldout_12",
|
| 118 |
+
"prompt": "Return only the integer: 737 + 103 - 43.",
|
| 119 |
+
"expected": "797",
|
| 120 |
+
"kind": "integer",
|
| 121 |
+
"output": "797",
|
| 122 |
+
"pass": true,
|
| 123 |
+
"tokens": 2
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"id": "heldout_13",
|
| 127 |
+
"prompt": "A crate has 740 red and 104 blue beads. Remove 44 red beads. Return only the total beads remaining.",
|
| 128 |
+
"expected": "800",
|
| 129 |
+
"kind": "integer",
|
| 130 |
+
"output": "676",
|
| 131 |
+
"pass": false,
|
| 132 |
+
"tokens": 2
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"id": "heldout_14",
|
| 136 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_743, count=105.",
|
| 137 |
+
"expected": {
|
| 138 |
+
"item": "part_743",
|
| 139 |
+
"count": 105
|
| 140 |
+
},
|
| 141 |
+
"kind": "json",
|
| 142 |
+
"output": "{\"item\": \"part_743\", \"count\": 105}",
|
| 143 |
+
"pass": true,
|
| 144 |
+
"tokens": 15
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"id": "heldout_15",
|
| 148 |
+
"prompt": "Return only one CSV value row, no header: name=Item746; quantity=106; code=K46.",
|
| 149 |
+
"expected": "Item746,106,K46",
|
| 150 |
+
"kind": "exact",
|
| 151 |
+
"output": "Item746,106,K46",
|
| 152 |
+
"pass": true,
|
| 153 |
+
"tokens": 7
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"id": "heldout_16",
|
| 157 |
+
"prompt": "Copy exactly, no other text: Key_749-B107_C47",
|
| 158 |
+
"expected": "Key_749-B107_C47",
|
| 159 |
+
"kind": "exact",
|
| 160 |
+
"output": "Key_749-B107_C47",
|
| 161 |
+
"pass": true,
|
| 162 |
+
"tokens": 8
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"id": "heldout_17",
|
| 166 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 108 + 31.",
|
| 167 |
+
"expected": "x * 108 + 31",
|
| 168 |
+
"kind": "code",
|
| 169 |
+
"output": "```python\ndef solve(x):\n return x * 108 + 31\n```",
|
| 170 |
+
"pass": true,
|
| 171 |
+
"tokens": 19
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"id": "heldout_18",
|
| 175 |
+
"prompt": "Return only the integer: 755 + 109 - 32.",
|
| 176 |
+
"expected": "832",
|
| 177 |
+
"kind": "integer",
|
| 178 |
+
"output": "832",
|
| 179 |
+
"pass": true,
|
| 180 |
+
"tokens": 2
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"id": "heldout_19",
|
| 184 |
+
"prompt": "A crate has 758 red and 110 blue beads. Remove 33 red beads. Return only the total beads remaining.",
|
| 185 |
+
"expected": "835",
|
| 186 |
+
"kind": "integer",
|
| 187 |
+
"output": "726",
|
| 188 |
+
"pass": false,
|
| 189 |
+
"tokens": 2
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"id": "heldout_20",
|
| 193 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_761, count=111.",
|
| 194 |
+
"expected": {
|
| 195 |
+
"item": "part_761",
|
| 196 |
+
"count": 111
|
| 197 |
+
},
|
| 198 |
+
"kind": "json",
|
| 199 |
+
"output": "{\"item\": \"part_761\", \"count\": 111}",
|
| 200 |
+
"pass": true,
|
| 201 |
+
"tokens": 15
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"id": "heldout_21",
|
| 205 |
+
"prompt": "Return only one CSV value row, no header: name=Item764; quantity=112; code=K35.",
|
| 206 |
+
"expected": "Item764,112,K35",
|
| 207 |
+
"kind": "exact",
|
| 208 |
+
"output": "Item764,112,K35",
|
| 209 |
+
"pass": true,
|
| 210 |
+
"tokens": 7
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"id": "heldout_22",
|
| 214 |
+
"prompt": "Copy exactly, no other text: Key_767-B113_C36",
|
| 215 |
+
"expected": "Key_767-B113_C36",
|
| 216 |
+
"kind": "exact",
|
| 217 |
+
"output": "Key_767-B113_C36",
|
| 218 |
+
"pass": true,
|
| 219 |
+
"tokens": 8
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"id": "heldout_23",
|
| 223 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 114 + 37.",
|
| 224 |
+
"expected": "x * 114 + 37",
|
| 225 |
+
"kind": "code",
|
| 226 |
+
"output": "```python\ndef solve(x):\n return x * 114 + 37\n```",
|
| 227 |
+
"pass": true,
|
| 228 |
+
"tokens": 19
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"id": "heldout_24",
|
| 232 |
+
"prompt": "Return only the integer: 773 + 115 - 38.",
|
| 233 |
+
"expected": "850",
|
| 234 |
+
"kind": "integer",
|
| 235 |
+
"output": "850",
|
| 236 |
+
"pass": true,
|
| 237 |
+
"tokens": 2
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"id": "heldout_25",
|
| 241 |
+
"prompt": "A crate has 776 red and 116 blue beads. Remove 39 red beads. Return only the total beads remaining.",
|
| 242 |
+
"expected": "853",
|
| 243 |
+
"kind": "integer",
|
| 244 |
+
"output": "726",
|
| 245 |
+
"pass": false,
|
| 246 |
+
"tokens": 2
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"id": "heldout_26",
|
| 250 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_779, count=117.",
|
| 251 |
+
"expected": {
|
| 252 |
+
"item": "part_779",
|
| 253 |
+
"count": 117
|
| 254 |
+
},
|
| 255 |
+
"kind": "json",
|
| 256 |
+
"output": "{\"item\": \"part_779\", \"count\": 117}",
|
| 257 |
+
"pass": true,
|
| 258 |
+
"tokens": 15
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"id": "heldout_27",
|
| 262 |
+
"prompt": "Return only one CSV value row, no header: name=Item782; quantity=118; code=K41.",
|
| 263 |
+
"expected": "Item782,118,K41",
|
| 264 |
+
"kind": "exact",
|
| 265 |
+
"output": "Item782,118,K41",
|
| 266 |
+
"pass": true,
|
| 267 |
+
"tokens": 7
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"id": "heldout_28",
|
| 271 |
+
"prompt": "Copy exactly, no other text: Key_785-B119_C42",
|
| 272 |
+
"expected": "Key_785-B119_C42",
|
| 273 |
+
"kind": "exact",
|
| 274 |
+
"output": "Key_785-B119_C42",
|
| 275 |
+
"pass": true,
|
| 276 |
+
"tokens": 8
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"id": "heldout_29",
|
| 280 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 120 + 43.",
|
| 281 |
+
"expected": "x * 120 + 43",
|
| 282 |
+
"kind": "code",
|
| 283 |
+
"output": "```python\ndef solve(x):\n return x * 120 + 43\n```",
|
| 284 |
+
"pass": true,
|
| 285 |
+
"tokens": 19
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"id": "heldout_30",
|
| 289 |
+
"prompt": "Return only the integer: 791 + 121 - 44.",
|
| 290 |
+
"expected": "868",
|
| 291 |
+
"kind": "integer",
|
| 292 |
+
"output": "868",
|
| 293 |
+
"pass": true,
|
| 294 |
+
"tokens": 2
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"id": "heldout_31",
|
| 298 |
+
"prompt": "A crate has 794 red and 122 blue beads. Remove 45 red beads. Return only the total beads remaining.",
|
| 299 |
+
"expected": "871",
|
| 300 |
+
"kind": "integer",
|
| 301 |
+
"output": "726",
|
| 302 |
+
"pass": false,
|
| 303 |
+
"tokens": 2
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"id": "heldout_32",
|
| 307 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_797, count=123.",
|
| 308 |
+
"expected": {
|
| 309 |
+
"item": "part_797",
|
| 310 |
+
"count": 123
|
| 311 |
+
},
|
| 312 |
+
"kind": "json",
|
| 313 |
+
"output": "{\"item\": \"part_797\", \"count\": 123}",
|
| 314 |
+
"pass": true,
|
| 315 |
+
"tokens": 15
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"id": "heldout_33",
|
| 319 |
+
"prompt": "Return only one CSV value row, no header: name=Item800; quantity=124; code=K47.",
|
| 320 |
+
"expected": "Item800,124,K47",
|
| 321 |
+
"kind": "exact",
|
| 322 |
+
"output": "Item800,124,K47",
|
| 323 |
+
"pass": true,
|
| 324 |
+
"tokens": 7
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"id": "heldout_34",
|
| 328 |
+
"prompt": "Copy exactly, no other text: Key_803-B125_C31",
|
| 329 |
+
"expected": "Key_803-B125_C31",
|
| 330 |
+
"kind": "exact",
|
| 331 |
+
"output": "Key_803-B125_C31",
|
| 332 |
+
"pass": true,
|
| 333 |
+
"tokens": 8
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"id": "heldout_35",
|
| 337 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 126 + 32.",
|
| 338 |
+
"expected": "x * 126 + 32",
|
| 339 |
+
"kind": "code",
|
| 340 |
+
"output": "```python\ndef solve(x):\n return x * 126 + 32\n```",
|
| 341 |
+
"pass": true,
|
| 342 |
+
"tokens": 19
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"id": "heldout_36",
|
| 346 |
+
"prompt": "Return only the integer: 809 + 127 - 33.",
|
| 347 |
+
"expected": "903",
|
| 348 |
+
"kind": "integer",
|
| 349 |
+
"output": "903",
|
| 350 |
+
"pass": true,
|
| 351 |
+
"tokens": 2
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"id": "heldout_37",
|
| 355 |
+
"prompt": "A crate has 812 red and 128 blue beads. Remove 34 red beads. Return only the total beads remaining.",
|
| 356 |
+
"expected": "906",
|
| 357 |
+
"kind": "integer",
|
| 358 |
+
"output": "812 - 34 = 778\nAnswer: 778",
|
| 359 |
+
"pass": false,
|
| 360 |
+
"tokens": 13
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"id": "heldout_38",
|
| 364 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_815, count=129.",
|
| 365 |
+
"expected": {
|
| 366 |
+
"item": "part_815",
|
| 367 |
+
"count": 129
|
| 368 |
+
},
|
| 369 |
+
"kind": "json",
|
| 370 |
+
"output": "{\"item\": \"part_815\", \"count\": 129}",
|
| 371 |
+
"pass": true,
|
| 372 |
+
"tokens": 15
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"id": "heldout_39",
|
| 376 |
+
"prompt": "Return only one CSV value row, no header: name=Item818; quantity=130; code=K36.",
|
| 377 |
+
"expected": "Item818,130,K36",
|
| 378 |
+
"kind": "exact",
|
| 379 |
+
"output": "Item818,130,K36",
|
| 380 |
+
"pass": true,
|
| 381 |
+
"tokens": 7
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"id": "heldout_40",
|
| 385 |
+
"prompt": "Copy exactly, no other text: Key_821-B131_C37",
|
| 386 |
+
"expected": "Key_821-B131_C37",
|
| 387 |
+
"kind": "exact",
|
| 388 |
+
"output": "Key_821-B131_C37",
|
| 389 |
+
"pass": true,
|
| 390 |
+
"tokens": 8
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"id": "heldout_41",
|
| 394 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 132 + 38.",
|
| 395 |
+
"expected": "x * 132 + 38",
|
| 396 |
+
"kind": "code",
|
| 397 |
+
"output": "```python\ndef solve(x):\n return x * 132 + 38\n```",
|
| 398 |
+
"pass": true,
|
| 399 |
+
"tokens": 19
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"id": "heldout_42",
|
| 403 |
+
"prompt": "Return only the integer: 827 + 133 - 39.",
|
| 404 |
+
"expected": "921",
|
| 405 |
+
"kind": "integer",
|
| 406 |
+
"output": "921",
|
| 407 |
+
"pass": true,
|
| 408 |
+
"tokens": 2
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"id": "heldout_43",
|
| 412 |
+
"prompt": "A crate has 830 red and 134 blue beads. Remove 40 red beads. Return only the total beads remaining.",
|
| 413 |
+
"expected": "924",
|
| 414 |
+
"kind": "integer",
|
| 415 |
+
"output": "830 - 40 = 790\nAnswer: 790",
|
| 416 |
+
"pass": false,
|
| 417 |
+
"tokens": 13
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"id": "heldout_44",
|
| 421 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_833, count=135.",
|
| 422 |
+
"expected": {
|
| 423 |
+
"item": "part_833",
|
| 424 |
+
"count": 135
|
| 425 |
+
},
|
| 426 |
+
"kind": "json",
|
| 427 |
+
"output": "{\"item\": \"part_833\", \"count\": 135}",
|
| 428 |
+
"pass": true,
|
| 429 |
+
"tokens": 15
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"id": "heldout_45",
|
| 433 |
+
"prompt": "Return only one CSV value row, no header: name=Item836; quantity=136; code=K42.",
|
| 434 |
+
"expected": "Item836,136,K42",
|
| 435 |
+
"kind": "exact",
|
| 436 |
+
"output": "Item836,136,K42",
|
| 437 |
+
"pass": true,
|
| 438 |
+
"tokens": 7
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"id": "heldout_46",
|
| 442 |
+
"prompt": "Copy exactly, no other text: Key_839-B137_C43",
|
| 443 |
+
"expected": "Key_839-B137_C43",
|
| 444 |
+
"kind": "exact",
|
| 445 |
+
"output": "Key_839-B137_C43",
|
| 446 |
+
"pass": true,
|
| 447 |
+
"tokens": 8
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"id": "heldout_47",
|
| 451 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 138 + 44.",
|
| 452 |
+
"expected": "x * 138 + 44",
|
| 453 |
+
"kind": "code",
|
| 454 |
+
"output": "```python\ndef solve(x):\n return x * 138 + 44\n```",
|
| 455 |
+
"pass": true,
|
| 456 |
+
"tokens": 19
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"id": "heldout_48",
|
| 460 |
+
"prompt": "Return only the integer: 845 + 139 - 45.",
|
| 461 |
+
"expected": "939",
|
| 462 |
+
"kind": "integer",
|
| 463 |
+
"output": "839",
|
| 464 |
+
"pass": false,
|
| 465 |
+
"tokens": 2
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"id": "heldout_49",
|
| 469 |
+
"prompt": "A crate has 848 red and 140 blue beads. Remove 46 red beads. Return only the total beads remaining.",
|
| 470 |
+
"expected": "942",
|
| 471 |
+
"kind": "integer",
|
| 472 |
+
"output": "847 are left.",
|
| 473 |
+
"pass": false,
|
| 474 |
+
"tokens": 5
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"id": "heldout_50",
|
| 478 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_851, count=141.",
|
| 479 |
+
"expected": {
|
| 480 |
+
"item": "part_851",
|
| 481 |
+
"count": 141
|
| 482 |
+
},
|
| 483 |
+
"kind": "json",
|
| 484 |
+
"output": "{\"item\": \"part_851\", \"count\": 141}",
|
| 485 |
+
"pass": true,
|
| 486 |
+
"tokens": 15
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"id": "heldout_51",
|
| 490 |
+
"prompt": "Return only one CSV value row, no header: name=Item854; quantity=142; code=K31.",
|
| 491 |
+
"expected": "Item854,142,K31",
|
| 492 |
+
"kind": "exact",
|
| 493 |
+
"output": "Item854,142,K31",
|
| 494 |
+
"pass": true,
|
| 495 |
+
"tokens": 7
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"id": "heldout_52",
|
| 499 |
+
"prompt": "Copy exactly, no other text: Key_857-B143_C32",
|
| 500 |
+
"expected": "Key_857-B143_C32",
|
| 501 |
+
"kind": "exact",
|
| 502 |
+
"output": "Key_857-B143_C32",
|
| 503 |
+
"pass": true,
|
| 504 |
+
"tokens": 8
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"id": "heldout_53",
|
| 508 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 144 + 33.",
|
| 509 |
+
"expected": "x * 144 + 33",
|
| 510 |
+
"kind": "code",
|
| 511 |
+
"output": "```python\ndef solve(x):\n return x * 144 + 33\n```",
|
| 512 |
+
"pass": true,
|
| 513 |
+
"tokens": 19
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"id": "heldout_54",
|
| 517 |
+
"prompt": "Return only the integer: 863 + 145 - 34.",
|
| 518 |
+
"expected": "974",
|
| 519 |
+
"kind": "integer",
|
| 520 |
+
"output": "974",
|
| 521 |
+
"pass": true,
|
| 522 |
+
"tokens": 2
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"id": "heldout_55",
|
| 526 |
+
"prompt": "A crate has 866 red and 146 blue beads. Remove 35 red beads. Return only the total beads remaining.",
|
| 527 |
+
"expected": "977",
|
| 528 |
+
"kind": "integer",
|
| 529 |
+
"output": "865 are left.",
|
| 530 |
+
"pass": false,
|
| 531 |
+
"tokens": 5
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"id": "heldout_56",
|
| 535 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_869, count=147.",
|
| 536 |
+
"expected": {
|
| 537 |
+
"item": "part_869",
|
| 538 |
+
"count": 147
|
| 539 |
+
},
|
| 540 |
+
"kind": "json",
|
| 541 |
+
"output": "{\"item\": \"part_869\", \"count\": 147}",
|
| 542 |
+
"pass": true,
|
| 543 |
+
"tokens": 15
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"id": "heldout_57",
|
| 547 |
+
"prompt": "Return only one CSV value row, no header: name=Item872; quantity=148; code=K37.",
|
| 548 |
+
"expected": "Item872,148,K37",
|
| 549 |
+
"kind": "exact",
|
| 550 |
+
"output": "Item872,148,K37",
|
| 551 |
+
"pass": true,
|
| 552 |
+
"tokens": 7
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"id": "heldout_58",
|
| 556 |
+
"prompt": "Copy exactly, no other text: Key_875-B149_C38",
|
| 557 |
+
"expected": "Key_875-B149_C38",
|
| 558 |
+
"kind": "exact",
|
| 559 |
+
"output": "Key_875-B149_C38",
|
| 560 |
+
"pass": true,
|
| 561 |
+
"tokens": 8
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"id": "heldout_59",
|
| 565 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 150 + 39.",
|
| 566 |
+
"expected": "x * 150 + 39",
|
| 567 |
+
"kind": "code",
|
| 568 |
+
"output": "```python\ndef solve(x):\n return x * 150 + 39\n```",
|
| 569 |
+
"pass": true,
|
| 570 |
+
"tokens": 19
|
| 571 |
+
}
|
| 572 |
+
]
|
v3-results/v3_reloaded_outputs.json
ADDED
|
@@ -0,0 +1,572 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"id": "heldout_00",
|
| 4 |
+
"prompt": "Return only the integer: 701 + 91 - 31.",
|
| 5 |
+
"expected": "761",
|
| 6 |
+
"kind": "integer",
|
| 7 |
+
"output": "761",
|
| 8 |
+
"pass": true,
|
| 9 |
+
"tokens": 2
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"id": "heldout_01",
|
| 13 |
+
"prompt": "A crate has 704 red and 92 blue beads. Remove 32 red beads. Return only the total beads remaining.",
|
| 14 |
+
"expected": "764",
|
| 15 |
+
"kind": "integer",
|
| 16 |
+
"output": "The number of red beads left is 672. The number of blue beads is still 92, for a total of 764 beads.",
|
| 17 |
+
"pass": false,
|
| 18 |
+
"tokens": 29
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"id": "heldout_02",
|
| 22 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_707, count=93.",
|
| 23 |
+
"expected": {
|
| 24 |
+
"item": "part_707",
|
| 25 |
+
"count": 93
|
| 26 |
+
},
|
| 27 |
+
"kind": "json",
|
| 28 |
+
"output": "{\"item\": \"part_707\", \"count\": 93}",
|
| 29 |
+
"pass": true,
|
| 30 |
+
"tokens": 15
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"id": "heldout_03",
|
| 34 |
+
"prompt": "Return only one CSV value row, no header: name=Item710; quantity=94; code=K34.",
|
| 35 |
+
"expected": "Item710,94,K34",
|
| 36 |
+
"kind": "exact",
|
| 37 |
+
"output": "Item710,94,K34",
|
| 38 |
+
"pass": true,
|
| 39 |
+
"tokens": 7
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"id": "heldout_04",
|
| 43 |
+
"prompt": "Copy exactly, no other text: Key_713-B95_C35",
|
| 44 |
+
"expected": "Key_713-B95_C35",
|
| 45 |
+
"kind": "exact",
|
| 46 |
+
"output": "Key_713-B95_C35",
|
| 47 |
+
"pass": true,
|
| 48 |
+
"tokens": 8
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"id": "heldout_05",
|
| 52 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 96 + 36.",
|
| 53 |
+
"expected": "x * 96 + 36",
|
| 54 |
+
"kind": "code",
|
| 55 |
+
"output": "```python\ndef solve(x):\n return x * 96 + 36\n```",
|
| 56 |
+
"pass": true,
|
| 57 |
+
"tokens": 19
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"id": "heldout_06",
|
| 61 |
+
"prompt": "Return only the integer: 719 + 97 - 37.",
|
| 62 |
+
"expected": "779",
|
| 63 |
+
"kind": "integer",
|
| 64 |
+
"output": "779",
|
| 65 |
+
"pass": true,
|
| 66 |
+
"tokens": 2
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"id": "heldout_07",
|
| 70 |
+
"prompt": "A crate has 722 red and 98 blue beads. Remove 38 red beads. Return only the total beads remaining.",
|
| 71 |
+
"expected": "782",
|
| 72 |
+
"kind": "integer",
|
| 73 |
+
"output": "The number of red beads left is 684. The number of beads left is 684 + 98 = 782.",
|
| 74 |
+
"pass": false,
|
| 75 |
+
"tokens": 26
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"id": "heldout_08",
|
| 79 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_725, count=99.",
|
| 80 |
+
"expected": {
|
| 81 |
+
"item": "part_725",
|
| 82 |
+
"count": 99
|
| 83 |
+
},
|
| 84 |
+
"kind": "json",
|
| 85 |
+
"output": "{\"item\": \"part_725\", \"count\": 99}",
|
| 86 |
+
"pass": true,
|
| 87 |
+
"tokens": 15
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"id": "heldout_09",
|
| 91 |
+
"prompt": "Return only one CSV value row, no header: name=Item728; quantity=100; code=K40.",
|
| 92 |
+
"expected": "Item728,100,K40",
|
| 93 |
+
"kind": "exact",
|
| 94 |
+
"output": "Item728,100,K40",
|
| 95 |
+
"pass": true,
|
| 96 |
+
"tokens": 7
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"id": "heldout_10",
|
| 100 |
+
"prompt": "Copy exactly, no other text: Key_731-B101_C41",
|
| 101 |
+
"expected": "Key_731-B101_C41",
|
| 102 |
+
"kind": "exact",
|
| 103 |
+
"output": "Key_731-B101_C41",
|
| 104 |
+
"pass": true,
|
| 105 |
+
"tokens": 8
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"id": "heldout_11",
|
| 109 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 102 + 42.",
|
| 110 |
+
"expected": "x * 102 + 42",
|
| 111 |
+
"kind": "code",
|
| 112 |
+
"output": "```python\ndef solve(x):\n return x * 102 + 42\n```",
|
| 113 |
+
"pass": true,
|
| 114 |
+
"tokens": 19
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"id": "heldout_12",
|
| 118 |
+
"prompt": "Return only the integer: 737 + 103 - 43.",
|
| 119 |
+
"expected": "797",
|
| 120 |
+
"kind": "integer",
|
| 121 |
+
"output": "797",
|
| 122 |
+
"pass": true,
|
| 123 |
+
"tokens": 2
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"id": "heldout_13",
|
| 127 |
+
"prompt": "A crate has 740 red and 104 blue beads. Remove 44 red beads. Return only the total beads remaining.",
|
| 128 |
+
"expected": "800",
|
| 129 |
+
"kind": "integer",
|
| 130 |
+
"output": "The number of red beads left is 696. The number of blue beads is unchanged at 104, for a total of 800 beads remaining.",
|
| 131 |
+
"pass": false,
|
| 132 |
+
"tokens": 31
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"id": "heldout_14",
|
| 136 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_743, count=105.",
|
| 137 |
+
"expected": {
|
| 138 |
+
"item": "part_743",
|
| 139 |
+
"count": 105
|
| 140 |
+
},
|
| 141 |
+
"kind": "json",
|
| 142 |
+
"output": "{\"item\": \"part_743\", \"count\": 105}",
|
| 143 |
+
"pass": true,
|
| 144 |
+
"tokens": 15
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"id": "heldout_15",
|
| 148 |
+
"prompt": "Return only one CSV value row, no header: name=Item746; quantity=106; code=K46.",
|
| 149 |
+
"expected": "Item746,106,K46",
|
| 150 |
+
"kind": "exact",
|
| 151 |
+
"output": "Item746,106,K46",
|
| 152 |
+
"pass": true,
|
| 153 |
+
"tokens": 7
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"id": "heldout_16",
|
| 157 |
+
"prompt": "Copy exactly, no other text: Key_749-B107_C47",
|
| 158 |
+
"expected": "Key_749-B107_C47",
|
| 159 |
+
"kind": "exact",
|
| 160 |
+
"output": "Key_749-B107_C47",
|
| 161 |
+
"pass": true,
|
| 162 |
+
"tokens": 8
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"id": "heldout_17",
|
| 166 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 108 + 31.",
|
| 167 |
+
"expected": "x * 108 + 31",
|
| 168 |
+
"kind": "code",
|
| 169 |
+
"output": "```python\ndef solve(x):\n return x * 108 + 31\n```",
|
| 170 |
+
"pass": true,
|
| 171 |
+
"tokens": 19
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"id": "heldout_18",
|
| 175 |
+
"prompt": "Return only the integer: 755 + 109 - 32.",
|
| 176 |
+
"expected": "832",
|
| 177 |
+
"kind": "integer",
|
| 178 |
+
"output": "732",
|
| 179 |
+
"pass": false,
|
| 180 |
+
"tokens": 2
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"id": "heldout_19",
|
| 184 |
+
"prompt": "A crate has 758 red and 110 blue beads. Remove 33 red beads. Return only the total beads remaining.",
|
| 185 |
+
"expected": "835",
|
| 186 |
+
"kind": "integer",
|
| 187 |
+
"output": "The number of red beads left is 758 - 33 = 725. The total number of beads left is 725 + 110 = 835.",
|
| 188 |
+
"pass": false,
|
| 189 |
+
"tokens": 33
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"id": "heldout_20",
|
| 193 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_761, count=111.",
|
| 194 |
+
"expected": {
|
| 195 |
+
"item": "part_761",
|
| 196 |
+
"count": 111
|
| 197 |
+
},
|
| 198 |
+
"kind": "json",
|
| 199 |
+
"output": "{\"item\": \"part_761\", \"count\": 111}",
|
| 200 |
+
"pass": true,
|
| 201 |
+
"tokens": 15
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"id": "heldout_21",
|
| 205 |
+
"prompt": "Return only one CSV value row, no header: name=Item764; quantity=112; code=K35.",
|
| 206 |
+
"expected": "Item764,112,K35",
|
| 207 |
+
"kind": "exact",
|
| 208 |
+
"output": "Item764,112,K35",
|
| 209 |
+
"pass": true,
|
| 210 |
+
"tokens": 7
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"id": "heldout_22",
|
| 214 |
+
"prompt": "Copy exactly, no other text: Key_767-B113_C36",
|
| 215 |
+
"expected": "Key_767-B113_C36",
|
| 216 |
+
"kind": "exact",
|
| 217 |
+
"output": "Key_767-B113_C36",
|
| 218 |
+
"pass": true,
|
| 219 |
+
"tokens": 8
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"id": "heldout_23",
|
| 223 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 114 + 37.",
|
| 224 |
+
"expected": "x * 114 + 37",
|
| 225 |
+
"kind": "code",
|
| 226 |
+
"output": "```python\ndef solve(x):\n return x * 114 + 37\n```",
|
| 227 |
+
"pass": true,
|
| 228 |
+
"tokens": 19
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"id": "heldout_24",
|
| 232 |
+
"prompt": "Return only the integer: 773 + 115 - 38.",
|
| 233 |
+
"expected": "850",
|
| 234 |
+
"kind": "integer",
|
| 235 |
+
"output": "750",
|
| 236 |
+
"pass": false,
|
| 237 |
+
"tokens": 2
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"id": "heldout_25",
|
| 241 |
+
"prompt": "A crate has 776 red and 116 blue beads. Remove 39 red beads. Return only the total beads remaining.",
|
| 242 |
+
"expected": "853",
|
| 243 |
+
"kind": "integer",
|
| 244 |
+
"output": "The number of red beads left is 737. There are 116 blue beads, 737 red beads, and 1533 beads left.",
|
| 245 |
+
"pass": false,
|
| 246 |
+
"tokens": 30
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"id": "heldout_26",
|
| 250 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_779, count=117.",
|
| 251 |
+
"expected": {
|
| 252 |
+
"item": "part_779",
|
| 253 |
+
"count": 117
|
| 254 |
+
},
|
| 255 |
+
"kind": "json",
|
| 256 |
+
"output": "{\"item\": \"part_779\", \"count\": 117}",
|
| 257 |
+
"pass": true,
|
| 258 |
+
"tokens": 15
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"id": "heldout_27",
|
| 262 |
+
"prompt": "Return only one CSV value row, no header: name=Item782; quantity=118; code=K41.",
|
| 263 |
+
"expected": "Item782,118,K41",
|
| 264 |
+
"kind": "exact",
|
| 265 |
+
"output": "Item782,118,K41",
|
| 266 |
+
"pass": true,
|
| 267 |
+
"tokens": 7
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"id": "heldout_28",
|
| 271 |
+
"prompt": "Copy exactly, no other text: Key_785-B119_C42",
|
| 272 |
+
"expected": "Key_785-B119_C42",
|
| 273 |
+
"kind": "exact",
|
| 274 |
+
"output": "Key_785-B119_C42",
|
| 275 |
+
"pass": true,
|
| 276 |
+
"tokens": 8
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"id": "heldout_29",
|
| 280 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 120 + 43.",
|
| 281 |
+
"expected": "x * 120 + 43",
|
| 282 |
+
"kind": "code",
|
| 283 |
+
"output": "```python\ndef solve(x):\n return x * 120 + 43\n```",
|
| 284 |
+
"pass": true,
|
| 285 |
+
"tokens": 19
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"id": "heldout_30",
|
| 289 |
+
"prompt": "Return only the integer: 791 + 121 - 44.",
|
| 290 |
+
"expected": "868",
|
| 291 |
+
"kind": "integer",
|
| 292 |
+
"output": "768",
|
| 293 |
+
"pass": false,
|
| 294 |
+
"tokens": 2
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"id": "heldout_31",
|
| 298 |
+
"prompt": "A crate has 794 red and 122 blue beads. Remove 45 red beads. Return only the total beads remaining.",
|
| 299 |
+
"expected": "871",
|
| 300 |
+
"kind": "integer",
|
| 301 |
+
"output": "The number of red beads left is 749 because 794 - 45 = 749. The total number of beads left is 749 + 122 = 871.",
|
| 302 |
+
"pass": false,
|
| 303 |
+
"tokens": 36
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"id": "heldout_32",
|
| 307 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_797, count=123.",
|
| 308 |
+
"expected": {
|
| 309 |
+
"item": "part_797",
|
| 310 |
+
"count": 123
|
| 311 |
+
},
|
| 312 |
+
"kind": "json",
|
| 313 |
+
"output": "{\"item\": \"part_797\", \"count\": 123}",
|
| 314 |
+
"pass": true,
|
| 315 |
+
"tokens": 15
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"id": "heldout_33",
|
| 319 |
+
"prompt": "Return only one CSV value row, no header: name=Item800; quantity=124; code=K47.",
|
| 320 |
+
"expected": "Item800,124,K47",
|
| 321 |
+
"kind": "exact",
|
| 322 |
+
"output": "Item800,124,K47",
|
| 323 |
+
"pass": true,
|
| 324 |
+
"tokens": 7
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"id": "heldout_34",
|
| 328 |
+
"prompt": "Copy exactly, no other text: Key_803-B125_C31",
|
| 329 |
+
"expected": "Key_803-B125_C31",
|
| 330 |
+
"kind": "exact",
|
| 331 |
+
"output": "Key_803-B125_C31",
|
| 332 |
+
"pass": true,
|
| 333 |
+
"tokens": 8
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"id": "heldout_35",
|
| 337 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 126 + 32.",
|
| 338 |
+
"expected": "x * 126 + 32",
|
| 339 |
+
"kind": "code",
|
| 340 |
+
"output": "```python\ndef solve(x):\n return x * 126 + 32\n```",
|
| 341 |
+
"pass": true,
|
| 342 |
+
"tokens": 19
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"id": "heldout_36",
|
| 346 |
+
"prompt": "Return only the integer: 809 + 127 - 33.",
|
| 347 |
+
"expected": "903",
|
| 348 |
+
"kind": "integer",
|
| 349 |
+
"output": "803",
|
| 350 |
+
"pass": false,
|
| 351 |
+
"tokens": 2
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"id": "heldout_37",
|
| 355 |
+
"prompt": "A crate has 812 red and 128 blue beads. Remove 34 red beads. Return only the total beads remaining.",
|
| 356 |
+
"expected": "906",
|
| 357 |
+
"kind": "integer",
|
| 358 |
+
"output": "The number of red beads left is 812 - 34 = 778. The total number of beads left is 778 + 128 = 906.",
|
| 359 |
+
"pass": false,
|
| 360 |
+
"tokens": 33
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"id": "heldout_38",
|
| 364 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_815, count=129.",
|
| 365 |
+
"expected": {
|
| 366 |
+
"item": "part_815",
|
| 367 |
+
"count": 129
|
| 368 |
+
},
|
| 369 |
+
"kind": "json",
|
| 370 |
+
"output": "{\"item\": \"part_815\", \"count\": 129}",
|
| 371 |
+
"pass": true,
|
| 372 |
+
"tokens": 15
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"id": "heldout_39",
|
| 376 |
+
"prompt": "Return only one CSV value row, no header: name=Item818; quantity=130; code=K36.",
|
| 377 |
+
"expected": "Item818,130,K36",
|
| 378 |
+
"kind": "exact",
|
| 379 |
+
"output": "Item818,130,K36",
|
| 380 |
+
"pass": true,
|
| 381 |
+
"tokens": 7
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"id": "heldout_40",
|
| 385 |
+
"prompt": "Copy exactly, no other text: Key_821-B131_C37",
|
| 386 |
+
"expected": "Key_821-B131_C37",
|
| 387 |
+
"kind": "exact",
|
| 388 |
+
"output": "Key_821-B131_C37",
|
| 389 |
+
"pass": true,
|
| 390 |
+
"tokens": 8
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"id": "heldout_41",
|
| 394 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 132 + 38.",
|
| 395 |
+
"expected": "x * 132 + 38",
|
| 396 |
+
"kind": "code",
|
| 397 |
+
"output": "```python\ndef solve(x):\n return x * 132 + 38\n```",
|
| 398 |
+
"pass": true,
|
| 399 |
+
"tokens": 19
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"id": "heldout_42",
|
| 403 |
+
"prompt": "Return only the integer: 827 + 133 - 39.",
|
| 404 |
+
"expected": "921",
|
| 405 |
+
"kind": "integer",
|
| 406 |
+
"output": "821",
|
| 407 |
+
"pass": false,
|
| 408 |
+
"tokens": 2
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"id": "heldout_43",
|
| 412 |
+
"prompt": "A crate has 830 red and 134 blue beads. Remove 40 red beads. Return only the total beads remaining.",
|
| 413 |
+
"expected": "924",
|
| 414 |
+
"kind": "integer",
|
| 415 |
+
"output": "The number of red beads left is 790 because 830 - 40 = 790. The total number of beads left is 790 + 134 = 924.",
|
| 416 |
+
"pass": false,
|
| 417 |
+
"tokens": 36
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"id": "heldout_44",
|
| 421 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_833, count=135.",
|
| 422 |
+
"expected": {
|
| 423 |
+
"item": "part_833",
|
| 424 |
+
"count": 135
|
| 425 |
+
},
|
| 426 |
+
"kind": "json",
|
| 427 |
+
"output": "{\"item\": \"part_833\", \"count\": 135}",
|
| 428 |
+
"pass": true,
|
| 429 |
+
"tokens": 15
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"id": "heldout_45",
|
| 433 |
+
"prompt": "Return only one CSV value row, no header: name=Item836; quantity=136; code=K42.",
|
| 434 |
+
"expected": "Item836,136,K42",
|
| 435 |
+
"kind": "exact",
|
| 436 |
+
"output": "Item836,136,K42",
|
| 437 |
+
"pass": true,
|
| 438 |
+
"tokens": 7
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"id": "heldout_46",
|
| 442 |
+
"prompt": "Copy exactly, no other text: Key_839-B137_C43",
|
| 443 |
+
"expected": "Key_839-B137_C43",
|
| 444 |
+
"kind": "exact",
|
| 445 |
+
"output": "Key_839-B137_C43",
|
| 446 |
+
"pass": true,
|
| 447 |
+
"tokens": 8
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"id": "heldout_47",
|
| 451 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 138 + 44.",
|
| 452 |
+
"expected": "x * 138 + 44",
|
| 453 |
+
"kind": "code",
|
| 454 |
+
"output": "```python\ndef solve(x):\n return x * 138 + 44\n```",
|
| 455 |
+
"pass": true,
|
| 456 |
+
"tokens": 19
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"id": "heldout_48",
|
| 460 |
+
"prompt": "Return only the integer: 845 + 139 - 45.",
|
| 461 |
+
"expected": "939",
|
| 462 |
+
"kind": "integer",
|
| 463 |
+
"output": "839",
|
| 464 |
+
"pass": false,
|
| 465 |
+
"tokens": 2
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"id": "heldout_49",
|
| 469 |
+
"prompt": "A crate has 848 red and 140 blue beads. Remove 46 red beads. Return only the total beads remaining.",
|
| 470 |
+
"expected": "942",
|
| 471 |
+
"kind": "integer",
|
| 472 |
+
"output": "The number of red beads left is 848 - 46 = 802. The total number of beads left is 802 + 140 = 942.",
|
| 473 |
+
"pass": false,
|
| 474 |
+
"tokens": 33
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"id": "heldout_50",
|
| 478 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_851, count=141.",
|
| 479 |
+
"expected": {
|
| 480 |
+
"item": "part_851",
|
| 481 |
+
"count": 141
|
| 482 |
+
},
|
| 483 |
+
"kind": "json",
|
| 484 |
+
"output": "{\"item\": \"part_851\", \"count\": 141}",
|
| 485 |
+
"pass": true,
|
| 486 |
+
"tokens": 15
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"id": "heldout_51",
|
| 490 |
+
"prompt": "Return only one CSV value row, no header: name=Item854; quantity=142; code=K31.",
|
| 491 |
+
"expected": "Item854,142,K31",
|
| 492 |
+
"kind": "exact",
|
| 493 |
+
"output": "Item854,142,K31",
|
| 494 |
+
"pass": true,
|
| 495 |
+
"tokens": 7
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"id": "heldout_52",
|
| 499 |
+
"prompt": "Copy exactly, no other text: Key_857-B143_C32",
|
| 500 |
+
"expected": "Key_857-B143_C32",
|
| 501 |
+
"kind": "exact",
|
| 502 |
+
"output": "Key_857-B143_C32",
|
| 503 |
+
"pass": true,
|
| 504 |
+
"tokens": 8
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"id": "heldout_53",
|
| 508 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 144 + 33.",
|
| 509 |
+
"expected": "x * 144 + 33",
|
| 510 |
+
"kind": "code",
|
| 511 |
+
"output": "```python\ndef solve(x):\n return x * 144 + 33\n```",
|
| 512 |
+
"pass": true,
|
| 513 |
+
"tokens": 19
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"id": "heldout_54",
|
| 517 |
+
"prompt": "Return only the integer: 863 + 145 - 34.",
|
| 518 |
+
"expected": "974",
|
| 519 |
+
"kind": "integer",
|
| 520 |
+
"output": "874",
|
| 521 |
+
"pass": false,
|
| 522 |
+
"tokens": 2
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"id": "heldout_55",
|
| 526 |
+
"prompt": "A crate has 866 red and 146 blue beads. Remove 35 red beads. Return only the total beads remaining.",
|
| 527 |
+
"expected": "977",
|
| 528 |
+
"kind": "integer",
|
| 529 |
+
"output": "The number of red beads left is 831. The number of blue beads is still 146, for a total of 977 beads.",
|
| 530 |
+
"pass": false,
|
| 531 |
+
"tokens": 29
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"id": "heldout_56",
|
| 535 |
+
"prompt": "Return only valid JSON with keys \"item\" and \"count\": item=part_869, count=147.",
|
| 536 |
+
"expected": {
|
| 537 |
+
"item": "part_869",
|
| 538 |
+
"count": 147
|
| 539 |
+
},
|
| 540 |
+
"kind": "json",
|
| 541 |
+
"output": "{\"item\": \"part_869\", \"count\": 147}",
|
| 542 |
+
"pass": true,
|
| 543 |
+
"tokens": 15
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"id": "heldout_57",
|
| 547 |
+
"prompt": "Return only one CSV value row, no header: name=Item872; quantity=148; code=K37.",
|
| 548 |
+
"expected": "Item872,148,K37",
|
| 549 |
+
"kind": "exact",
|
| 550 |
+
"output": "Item872,148,K37",
|
| 551 |
+
"pass": true,
|
| 552 |
+
"tokens": 7
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"id": "heldout_58",
|
| 556 |
+
"prompt": "Copy exactly, no other text: Key_875-B149_C38",
|
| 557 |
+
"expected": "Key_875-B149_C38",
|
| 558 |
+
"kind": "exact",
|
| 559 |
+
"output": "Key_875-B149_C38",
|
| 560 |
+
"pass": true,
|
| 561 |
+
"tokens": 8
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"id": "heldout_59",
|
| 565 |
+
"prompt": "Write only a Python function named solve(x) that returns x * 150 + 39.",
|
| 566 |
+
"expected": "x * 150 + 39",
|
| 567 |
+
"kind": "code",
|
| 568 |
+
"output": "```python\ndef solve(x):\n return x * 150 + 39\n```",
|
| 569 |
+
"pass": true,
|
| 570 |
+
"tokens": 19
|
| 571 |
+
}
|
| 572 |
+
]
|
v3-results/v3_report.json
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"parent_revision": "d9aa899722be4788f57107c2951c72d0ec6614ee",
|
| 4 |
+
"hardware": "NVIDIA A10G",
|
| 5 |
+
"job_timeout_seconds": 5400,
|
| 6 |
+
"listed_usd_per_hour": 1.0,
|
| 7 |
+
"training_seed": 20261113,
|
| 8 |
+
"baseline_passed": 40,
|
| 9 |
+
"elapsed_seconds": 622.09,
|
| 10 |
+
"training_examples": 1713,
|
| 11 |
+
"trainable_parameters": 41943040,
|
| 12 |
+
"training": {
|
| 13 |
+
"train_runtime": 390.2404,
|
| 14 |
+
"train_samples_per_second": 1.845,
|
| 15 |
+
"train_steps_per_second": 0.461,
|
| 16 |
+
"total_flos": 3960249019342848.0,
|
| 17 |
+
"train_loss": 0.2867420093880759,
|
| 18 |
+
"epoch": 0.4203152364273205
|
| 19 |
+
},
|
| 20 |
+
"optimizer_steps": 180,
|
| 21 |
+
"candidate_passed": 49,
|
| 22 |
+
"reloaded_passed": 43,
|
| 23 |
+
"groups": {
|
| 24 |
+
"integer": {
|
| 25 |
+
"baseline": 0,
|
| 26 |
+
"candidate": 3,
|
| 27 |
+
"total": 20
|
| 28 |
+
},
|
| 29 |
+
"exact": {
|
| 30 |
+
"baseline": 20,
|
| 31 |
+
"candidate": 20,
|
| 32 |
+
"total": 20
|
| 33 |
+
},
|
| 34 |
+
"code": {
|
| 35 |
+
"baseline": 10,
|
| 36 |
+
"candidate": 10,
|
| 37 |
+
"total": 10
|
| 38 |
+
},
|
| 39 |
+
"json": {
|
| 40 |
+
"baseline": 10,
|
| 41 |
+
"candidate": 10,
|
| 42 |
+
"total": 10
|
| 43 |
+
}
|
| 44 |
+
},
|
| 45 |
+
"quality_gate_passed": true,
|
| 46 |
+
"peak_gpu_bytes": 18991949312
|
| 47 |
+
}
|
v3/independent_eval.py
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# /// script
|
| 2 |
+
# dependencies = ["torch==2.8.0", "transformers==4.56.2", "huggingface_hub==0.34.4", "accelerate==1.10.1", "safetensors==0.6.2"]
|
| 3 |
+
# ///
|
| 4 |
+
"""Independent GSM8K and HumanEval subset comparison on remote GPUs."""
|
| 5 |
+
import json,re,sys,subprocess,tempfile,ast,urllib.request,urllib.parse,gc,hashlib
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
import torch
|
| 8 |
+
from transformers import AutoModelForCausalLM,AutoTokenizer
|
| 9 |
+
from huggingface_hub import HfApi,hf_hub_download,CommitOperationAdd
|
| 10 |
+
def task_check(case,text):
|
| 11 |
+
final=text.split("</think>")[-1]
|
| 12 |
+
numbers=re.findall(r"-?\d+(?:\.\d+)?",final.replace(",",""))
|
| 13 |
+
return bool(numbers) and abs(float(numbers[-1])-float(case["expected"]))<1e-6
|
| 14 |
+
api=HfApi();REPO="SEE7/Astheral";BASE="NousResearch/DeepHermes-3-Llama-3-8B-Preview";REV="53d902c7dcfa4c749a3b455149064f82af2e2549"
|
| 15 |
+
ROOT=Path(__file__).resolve().parent
|
| 16 |
+
BRANCH="v3-candidate-20261113"
|
| 17 |
+
revision=api.model_info(REPO,revision=BRANCH).sha
|
| 18 |
+
settings=json.loads(Path(hf_hub_download(REPO,"settings.json",revision=revision)).read_text())
|
| 19 |
+
def rows(repo,config,split,offset,length):
|
| 20 |
+
q=urllib.parse.urlencode(dict(dataset=repo,config=config,split=split,offset=offset,length=length))
|
| 21 |
+
return [r["row"] for r in json.load(urllib.request.urlopen("https://datasets-server.huggingface.co/rows?"+q,timeout=60))["rows"]]
|
| 22 |
+
math_revision=api.dataset_info("openai/gsm8k").sha
|
| 23 |
+
code_revision=api.dataset_info("openai/openai_humaneval").sha
|
| 24 |
+
cases=[]
|
| 25 |
+
for i,r in enumerate(rows("openai/gsm8k","main","test",231,8)):
|
| 26 |
+
expected=float(r["answer"].rsplit("####",1)[1].strip().replace(",",""))
|
| 27 |
+
cases.append({"id":"gsm8k-test-"+str(231+i),"mode":"thinking","prompt":r["question"]+"\nGive the final numerical answer after your explanation.","kind":"number","expected":expected,"source":"openai/gsm8k","source_revision":math_revision})
|
| 28 |
+
chosen={"HumanEval/"+str(i) for i in [0,3,7,12,18,25,36,45]}
|
| 29 |
+
for r in rows("openai/openai_humaneval","openai_humaneval","test",0,50):
|
| 30 |
+
if r["task_id"] not in chosen:continue
|
| 31 |
+
cases.append({"id":r["task_id"],"mode":"instruct","prompt":"Return only the complete Python implementation, including necessary imports, for the following function specification:\n\n"+r["prompt"],"kind":"code","entry_point":r["entry_point"],"tests":r["test"]+"\ncheck("+r["entry_point"]+")","source":"openai/openai_humaneval","source_revision":code_revision})
|
| 32 |
+
assert len(cases)==16
|
| 33 |
+
def check_code(c,text):
|
| 34 |
+
blocks=re.findall(r"\x60\x60\x60(?:python)?\s*\n(.*?)\x60\x60\x60",text,re.S)
|
| 35 |
+
code=blocks[0] if blocks else text.strip()
|
| 36 |
+
try:
|
| 37 |
+
tree=ast.parse(code)
|
| 38 |
+
for n in ast.walk(tree):
|
| 39 |
+
if isinstance(n,(ast.Import,ast.ImportFrom)):
|
| 40 |
+
names=[a.name.split(".")[0] for a in n.names] if isinstance(n,ast.Import) else [(n.module or "").split(".")[0]]
|
| 41 |
+
if any(x not in ["math","collections","itertools","functools","re","typing","statistics","heapq","bisect","string"] for x in names):return False
|
| 42 |
+
if isinstance(n,ast.Name) and n.id in ["open","exec","eval","compile","__import__","__builtins__","globals","locals","getattr","setattr","input"]:return False
|
| 43 |
+
if isinstance(n,ast.Attribute) and n.attr.startswith("__"):return False
|
| 44 |
+
with tempfile.TemporaryDirectory() as d:
|
| 45 |
+
p=Path(d)/"check.py"
|
| 46 |
+
p.write_text("import resource\nresource.setrlimit(resource.RLIMIT_CPU,(3,3))\nresource.setrlimit(resource.RLIMIT_AS,(536870912,536870912))\n"+code+"\n"+c["tests"])
|
| 47 |
+
r=subprocess.run([sys.executable,"-I",str(p)],cwd=d,timeout=5,capture_output=True,env={"PATH":"/usr/bin:/bin"})
|
| 48 |
+
return r.returncode==0
|
| 49 |
+
except Exception:return False
|
| 50 |
+
all_results={};scores={};paired_results={}
|
| 51 |
+
source=Path(hf_hub_download(REPO,'v3/train_v3.py',revision=revision)).read_text()
|
| 52 |
+
grade_node=next(n for n in ast.parse(source).body if isinstance(n,ast.FunctionDef) and n.name=='grade')
|
| 53 |
+
grade_scope={'json':json,'re':re};exec(compile(ast.Module(body=[grade_node],type_ignores=[]),'grader','exec'),grade_scope)
|
| 54 |
+
paired_cases=json.loads(Path(hf_hub_download(REPO,'v3-results/heldout_cases.json',revision=revision)).read_text())
|
| 55 |
+
for tag,repo,rev in [("base",REPO,"d9aa899722be4788f57107c2951c72d0ec6614ee"),("astheral",REPO,revision)]:
|
| 56 |
+
tok=AutoTokenizer.from_pretrained(repo,revision=rev);tok.pad_token_id=128004;tok.padding_side="left"
|
| 57 |
+
model=AutoModelForCausalLM.from_pretrained(repo,revision=rev,torch_dtype=torch.bfloat16,device_map={"":"cuda:0"},attn_implementation="sdpa").eval()
|
| 58 |
+
paired=[]
|
| 59 |
+
with torch.inference_mode():
|
| 60 |
+
for c in paired_cases:
|
| 61 |
+
inp=tok.apply_chat_template([{"role":"system","content":settings["system"]},{"role":"user","content":c["prompt"]}],tokenize=True,add_generation_prompt=True,return_dict=True,return_tensors="pt").to("cuda")
|
| 62 |
+
output=model.generate(**inp,max_new_tokens=96,do_sample=False,pad_token_id=128004,eos_token_id=128009)
|
| 63 |
+
text=tok.decode(output[0,inp["input_ids"].shape[1]:],skip_special_tokens=True)
|
| 64 |
+
paired.append({**c,"output":text,"pass":grade_scope["grade"](c,text)})
|
| 65 |
+
print("BF16_PAIRED",tag,c["id"],paired[-1]["pass"],flush=True)
|
| 66 |
+
paired_results[tag]=paired
|
| 67 |
+
result=[]
|
| 68 |
+
with torch.inference_mode():
|
| 69 |
+
for mode in ["instruct","thinking"]:
|
| 70 |
+
selected=[c for c in cases if c["mode"]==mode]
|
| 71 |
+
for i in range(0,len(selected),2):
|
| 72 |
+
group=selected[i:i+2]
|
| 73 |
+
system=settings["system"]+("\n"+settings["thinking"] if mode=="thinking" else "")
|
| 74 |
+
texts=[tok.apply_chat_template([{"role":"system","content":system},{"role":"user","content":c["prompt"]}],tokenize=False,add_generation_prompt=True) for c in group]
|
| 75 |
+
inputs=tok(texts,padding=True,return_tensors="pt").to("cuda")
|
| 76 |
+
out=model.generate(**inputs,max_new_tokens=1536 if mode=="thinking" else 768,do_sample=False,pad_token_id=128004,eos_token_id=128009)
|
| 77 |
+
for c,o in zip(group,out):
|
| 78 |
+
ids=o[inputs["input_ids"].shape[1]:]
|
| 79 |
+
text=tok.decode(ids,skip_special_tokens=True)
|
| 80 |
+
passed=check_code(c,text) if mode=="instruct" else task_check(c,text)
|
| 81 |
+
result.append({**c,"output":text,"pass":passed,"hit_token_limit":len(ids)>=(1536 if mode=="thinking" else 768) and 128009 not in ids.tolist()})
|
| 82 |
+
print("HOLDOUT",tag,c["id"],passed,flush=True)
|
| 83 |
+
all_results[tag]=result
|
| 84 |
+
scores[tag]={}
|
| 85 |
+
for kind in ["code","number"]:
|
| 86 |
+
chosen_rows=[r for r in result if r["kind"]==kind]
|
| 87 |
+
scores[tag][kind]={"passed":sum(r["pass"] for r in chosen_rows),"total":len(chosen_rows),"token_limited":sum(r["hit_token_limit"] for r in chosen_rows)}
|
| 88 |
+
del model;gc.collect();torch.cuda.empty_cache()
|
| 89 |
+
passed=sum(v["passed"] for v in scores["astheral"].values())>=sum(v["passed"] for v in scores["base"].values()) and all(scores["astheral"][k]["passed"]>=scores["base"][k]["passed"] for k in ["code","number"])
|
| 90 |
+
passed=passed and sum(r["pass"] for r in paired_results["astheral"])>=sum(r["pass"] for r in paired_results["base"])
|
| 91 |
+
report={"comparison":"V2 versus V3, both BF16","v2_revision":"d9aa899722be4788f57107c2951c72d0ec6614ee","paired_bf16_scores":{k:sum(r["pass"] for r in v) for k,v in paired_results.items()},"suite":"GSM8K test rows 231-238; HumanEval selected tasks 0,3,7,12,18,25,36,45","scores":scores,"no_material_regression_gate":passed,"astheral_revision":revision,"settings":{"sampling":False,"code_max_new_tokens":768,"math_max_new_tokens":1536},"hardware":torch.cuda.get_device_name(0),"source_revisions":{"gsm8k":math_revision,"humaneval":code_revision}}
|
| 92 |
+
changed={"paired_bf16_outputs.json":json.dumps(paired_results,indent=2),"holdout_evaluation.json":json.dumps(report,indent=2),"holdout_outputs.json":json.dumps(all_results,indent=2),"holdout_eval.py":Path(__file__).read_text(),"HUMANEVAL_LICENSE.txt":urllib.request.urlopen("https://raw.githubusercontent.com/openai/human-eval/master/LICENSE",timeout=60).read().decode()}
|
| 93 |
+
manifest=json.loads(Path(hf_hub_download(REPO,"artifact_manifest.json")).read_text())
|
| 94 |
+
for n,b in changed.items():manifest[n]={"bytes":len(b.encode()),"sha256":hashlib.sha256(b.encode()).hexdigest()}
|
| 95 |
+
changed["artifact_manifest.json"]=json.dumps(manifest,indent=2)
|
| 96 |
+
api.create_commit(repo_id=REPO,revision=BRANCH,operations=[CommitOperationAdd(path_in_repo=n,path_or_fileobj=b.encode()) for n,b in changed.items()],commit_message="Record measured V2 versus V3 GSM8K and HumanEval comparison")
|
| 97 |
+
print("HOLDOUT_REPORT",json.dumps(report),flush=True)
|
| 98 |
+
if not passed:raise SystemExit(5)
|
v3/train_v3.py
ADDED
|
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# /// script
|
| 2 |
+
# dependencies = ["torch==2.8.0", "transformers==4.56.2", "peft==0.17.1", "accelerate==1.10.1", "huggingface_hub==0.34.4", "bitsandbytes==0.47.0", "safetensors==0.6.2"]
|
| 3 |
+
# ///
|
| 4 |
+
"""Budget-bounded V3 training with paired, disjoint evaluation and reload gate."""
|
| 5 |
+
import os, json, random, time, re, gc, hashlib, traceback
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
import torch
|
| 8 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, Trainer, TrainingArguments, TrainerCallback, set_seed
|
| 9 |
+
from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training, PeftModel
|
| 10 |
+
from huggingface_hub import HfApi, hf_hub_download
|
| 11 |
+
|
| 12 |
+
REPO='SEE7/Astheral'; REV='d9aa899722be4788f57107c2951c72d0ec6614ee'
|
| 13 |
+
OUT=Path('/tmp/astheral-v3'); OUT.mkdir(exist_ok=True)
|
| 14 |
+
api=HfApi(); started=time.monotonic(); set_seed(20261113); rng=random.Random(20261113)
|
| 15 |
+
assert torch.cuda.is_available() and 'A10G' in torch.cuda.get_device_name(0)
|
| 16 |
+
branch='v3-candidate-20261113'
|
| 17 |
+
api.create_branch(REPO,branch=branch,revision=REV,exist_ok=True)
|
| 18 |
+
settings=json.loads(Path(hf_hub_download(REPO,'settings.json',revision=REV)).read_text())
|
| 19 |
+
tok=AutoTokenizer.from_pretrained(REPO,revision=REV); tok.pad_token_id=128004
|
| 20 |
+
report={'status':'preparing','parent_revision':REV,'hardware':torch.cuda.get_device_name(0),'job_timeout_seconds':5400,'listed_usd_per_hour':1.0,'training_seed':20261113}
|
| 21 |
+
|
| 22 |
+
def publish():
|
| 23 |
+
report['elapsed_seconds']=round(time.monotonic()-started,2)
|
| 24 |
+
(OUT/'v3_report.json').write_text(json.dumps(report,indent=2))
|
| 25 |
+
api.upload_folder(repo_id=REPO,folder_path=OUT,revision=branch,path_in_repo='v3-results',ignore_patterns=['merged/**'],commit_message='Record measured V3 training and evaluation artifacts')
|
| 26 |
+
|
| 27 |
+
def sample(a,b,c,kind):
|
| 28 |
+
if kind==0:return f'Return only the integer: {a} + {b} - {c}.',str(a+b-c),'integer'
|
| 29 |
+
if kind==1:return f'A crate has {a} red and {b} blue beads. Remove {c} red beads. Return only the total beads remaining.',str(a+b-c),'integer'
|
| 30 |
+
if kind==2:return f'Return only valid JSON with keys "item" and "count": item=part_{a}, count={b}.',{'item':f'part_{a}','count':b},'json'
|
| 31 |
+
if kind==3:return f'Return only one CSV value row, no header: name=Item{a}; quantity={b}; code=K{c}.',f'Item{a},{b},K{c}','exact'
|
| 32 |
+
if kind==4:return f'Copy exactly, no other text: Key_{a}-B{b}_C{c}',f'Key_{a}-B{b}_C{c}','exact'
|
| 33 |
+
return f'Write only a Python function named solve(x) that returns x * {b} + {c}.',f'x * {b} + {c}','code'
|
| 34 |
+
|
| 35 |
+
def grade(case,text):
|
| 36 |
+
expected=case['expected']; kind=case['kind']; text=text.strip()
|
| 37 |
+
try:
|
| 38 |
+
if kind=='json':return json.loads(text)==expected
|
| 39 |
+
if kind=='integer':return re.fullmatch(r'-?\d+',text) is not None and int(text)==int(expected)
|
| 40 |
+
if kind=='exact':return text==expected
|
| 41 |
+
# Parse and evaluate a tiny arithmetic-only AST; never execute model-generated code.
|
| 42 |
+
import ast
|
| 43 |
+
clean=re.sub(r'^```(?:python)?\s*|\s*```$','',text).strip();tree=ast.parse(clean)
|
| 44 |
+
if len(tree.body)!=1 or not isinstance(tree.body[0],ast.FunctionDef):return False
|
| 45 |
+
fn=tree.body[0]
|
| 46 |
+
if fn.name!='solve' or len(fn.args.args)!=1 or fn.args.args[0].arg!='x' or len(fn.body)!=1 or not isinstance(fn.body[0],ast.Return):return False
|
| 47 |
+
def calc(n,x):
|
| 48 |
+
if isinstance(n,ast.Constant) and isinstance(n.value,(int,float)):return n.value
|
| 49 |
+
if isinstance(n,ast.Name) and n.id=='x':return x
|
| 50 |
+
if isinstance(n,ast.BinOp) and isinstance(n.op,(ast.Add,ast.Sub,ast.Mult)):
|
| 51 |
+
a,b=calc(n.left,x),calc(n.right,x)
|
| 52 |
+
return a+b if isinstance(n.op,ast.Add) else a-b if isinstance(n.op,ast.Sub) else a*b
|
| 53 |
+
raise ValueError('Unsupported expression')
|
| 54 |
+
match=re.fullmatch(r'x \* (\d+) \+ (\d+)',expected);b,c=map(int,match.groups())
|
| 55 |
+
return all(calc(fn.body[0].value,x)==x*b+c for x in [-5,0,1,7,100])
|
| 56 |
+
except Exception:return False
|
| 57 |
+
|
| 58 |
+
tests=[]
|
| 59 |
+
for i in range(60):
|
| 60 |
+
# Evaluation numbers are outside the synthetic training ranges.
|
| 61 |
+
a,b,c=701+i*3,91+i,31+i%17;p,e,k=sample(a,b,c,i%6)
|
| 62 |
+
tests.append({'id':f'heldout_{i:02}','prompt':p,'expected':e,'kind':k})
|
| 63 |
+
train=[]
|
| 64 |
+
for i in range(900):
|
| 65 |
+
a,b,c=rng.randint(100,500),rng.randint(10,80),rng.randint(1,29)
|
| 66 |
+
p,e,k=sample(a,b,c,i%6)
|
| 67 |
+
answer=json.dumps(e) if k=='json' else 'def solve(x):\n return '+e if k=='code' else str(e)
|
| 68 |
+
train.append({'messages':[{'role':'user','content':p},{'role':'assistant','content':answer}]})
|
| 69 |
+
retained=[json.loads(s) for s in Path(hf_hub_download(REPO,'train.jsonl',revision=REV)).read_text().splitlines()]
|
| 70 |
+
rng.shuffle(retained);train+=retained[:900];rng.shuffle(train)
|
| 71 |
+
assert not {x['prompt'] for x in tests} & {m['content'] for row in train for m in row['messages'] if m['role']=='user'}
|
| 72 |
+
(OUT/'heldout_cases.json').write_text(json.dumps(tests,indent=2));(OUT/'training.jsonl').write_text('\n'.join(json.dumps(x) for x in train)+'\n')
|
| 73 |
+
(OUT/'train_v3.py').write_text(Path(__file__).read_text())
|
| 74 |
+
q=BitsAndBytesConfig(load_in_4bit=True,bnb_4bit_quant_type='nf4',bnb_4bit_use_double_quant=True,bnb_4bit_compute_dtype=torch.bfloat16)
|
| 75 |
+
model=AutoModelForCausalLM.from_pretrained(REPO,revision=REV,quantization_config=q,device_map={'':'cuda:0'},torch_dtype=torch.bfloat16,attn_implementation='sdpa')
|
| 76 |
+
|
| 77 |
+
@torch.inference_mode()
|
| 78 |
+
def evaluate(label):
|
| 79 |
+
model.eval(); rows=[]
|
| 80 |
+
for i,case in enumerate(tests):
|
| 81 |
+
inputs=tok.apply_chat_template([{'role':'system','content':settings['system']},{'role':'user','content':case['prompt']}],tokenize=True,add_generation_prompt=True,return_dict=True,return_tensors='pt').to('cuda')
|
| 82 |
+
out=model.generate(**inputs,max_new_tokens=96,do_sample=False,pad_token_id=128004,eos_token_id=128009)
|
| 83 |
+
text=tok.decode(out[0,inputs['input_ids'].shape[1]:],skip_special_tokens=True)
|
| 84 |
+
rows.append({**case,'output':text,'pass':grade(case,text),'tokens':int(out.shape[1]-inputs['input_ids'].shape[1])})
|
| 85 |
+
print('EVAL',label,i,rows[-1]['pass'],flush=True)
|
| 86 |
+
(OUT/(label+'_outputs.json')).write_text(json.dumps(rows,indent=2))
|
| 87 |
+
return rows
|
| 88 |
+
|
| 89 |
+
try:
|
| 90 |
+
baseline=evaluate('v2_quantized'); report['baseline_passed']=sum(r['pass'] for r in baseline);publish()
|
| 91 |
+
model=prepare_model_for_kbit_training(model,use_gradient_checkpointing=True)
|
| 92 |
+
model=get_peft_model(model,LoraConfig(r=16,lora_alpha=32,lora_dropout=0.05,target_modules=['q_proj','k_proj','v_proj','o_proj','gate_proj','up_proj','down_proj'],task_type='CAUSAL_LM'))
|
| 93 |
+
model.config.use_cache=False;encoded=[]
|
| 94 |
+
for row in train:
|
| 95 |
+
messages=row['messages']; prefix=tok.apply_chat_template(messages[:-1],tokenize=True,add_generation_prompt=True)
|
| 96 |
+
full=tok.apply_chat_template(messages,tokenize=True,add_generation_prompt=False)
|
| 97 |
+
if len(full)<=512 and full[:len(prefix)]==prefix:encoded.append({'input_ids':full,'labels':[-100]*len(prefix)+full[len(prefix):]})
|
| 98 |
+
def collate(rows):
|
| 99 |
+
length=max(len(r['input_ids']) for r in rows)
|
| 100 |
+
return {k:torch.tensor([r[k]+[pad]*(length-len(r[k])) for r in rows]) for k,pad in [('input_ids',128004),('labels',-100)]}|{'attention_mask':torch.tensor([[1]*len(r['input_ids'])+[0]*(length-len(r['input_ids'])) for r in rows])}
|
| 101 |
+
class Budget(TrainerCallback):
|
| 102 |
+
def on_step_end(self,args,state,control,**kwargs):
|
| 103 |
+
if time.monotonic()-started>2400:control.should_training_stop=True
|
| 104 |
+
return control
|
| 105 |
+
args=TrainingArguments(output_dir='/tmp/v3-checkpoints',max_steps=180,per_device_train_batch_size=1,gradient_accumulation_steps=4,learning_rate=1e-5,warmup_ratio=0.05,bf16=True,gradient_checkpointing=True,gradient_checkpointing_kwargs={'use_reentrant':False},optim='paged_adamw_8bit',save_strategy='no',logging_steps=10,report_to='none',remove_unused_columns=False,seed=20261113)
|
| 106 |
+
trainer=Trainer(model=model,args=args,train_dataset=encoded,data_collator=collate,callbacks=[Budget()])
|
| 107 |
+
report.update(status='training',training_examples=len(encoded),trainable_parameters=sum(p.numel() for p in model.parameters() if p.requires_grad));publish()
|
| 108 |
+
result=trainer.train();report['training']=result.metrics;report['optimizer_steps']=trainer.state.global_step
|
| 109 |
+
adapter=OUT/'adapter';model.save_pretrained(adapter,safe_serialization=True);tok.save_pretrained(adapter)
|
| 110 |
+
model.config.use_cache=True;model.gradient_checkpointing_disable();candidate=evaluate('v3_quantized')
|
| 111 |
+
report['candidate_passed']=sum(r['pass'] for r in candidate);report['status']='trained';publish()
|
| 112 |
+
del trainer,model;gc.collect();torch.cuda.empty_cache()
|
| 113 |
+
model=AutoModelForCausalLM.from_pretrained(REPO,revision=REV,torch_dtype=torch.bfloat16,device_map={'':'cuda:0'},attn_implementation='sdpa')
|
| 114 |
+
model=PeftModel.from_pretrained(model,adapter).merge_and_unload(safe_merge=True)
|
| 115 |
+
model.config.astheral_version='3';merged=OUT/'merged';model.save_pretrained(merged,safe_serialization=True,max_shard_size='4GB');tok.save_pretrained(merged)
|
| 116 |
+
del model;gc.collect();torch.cuda.empty_cache()
|
| 117 |
+
model=AutoModelForCausalLM.from_pretrained(merged,torch_dtype=torch.bfloat16,device_map={'':'cuda:0'},attn_implementation='sdpa')
|
| 118 |
+
reloaded=evaluate('v3_reloaded');report['reloaded_passed']=sum(r['pass'] for r in reloaded)
|
| 119 |
+
groups={kind:{'baseline':sum(r['pass'] for r in baseline if r['kind']==kind),'candidate':sum(r['pass'] for r in reloaded if r['kind']==kind),'total':sum(r['kind']==kind for r in tests)} for kind in {x['kind'] for x in tests}}
|
| 120 |
+
report['groups']=groups;report['quality_gate_passed']=report['reloaded_passed']>report['baseline_passed'] and all(v['candidate']>=v['baseline'] for v in groups.values())
|
| 121 |
+
report['status']='complete';report['peak_gpu_bytes']=torch.cuda.max_memory_allocated();publish()
|
| 122 |
+
# Save the actual trained weights to the isolated candidate branch even on a failed gate.
|
| 123 |
+
api.upload_folder(repo_id=REPO,folder_path=merged,revision=branch,commit_message='Save reloaded V3 candidate weights; see measured quality gate in v3-results')
|
| 124 |
+
if report['quality_gate_passed']:
|
| 125 |
+
api.create_tag(REPO,tag='v3',revision=api.model_info(REPO,revision=branch).sha,exist_ok=True)
|
| 126 |
+
print('V3_COMPLETE',json.dumps(report),flush=True)
|
| 127 |
+
except Exception:
|
| 128 |
+
report['status']='failed';report['error']=traceback.format_exc();publish();raise
|