初始化项目，由ModelHub XC社区提供模型
Model: rita-cohere/tya-eng-v1 Source: Original Platform
This commit is contained in:
ModelHub XC
50
.gitattributes
vendored
Normal file
50
.gitattributes
vendored
Normal file
@@ -0,0 +1,50 @@
|
||||
*.7z filter=lfs diff=lfs merge=lfs -text
|
||||
*.arrow filter=lfs diff=lfs merge=lfs -text
|
||||
*.bin filter=lfs diff=lfs merge=lfs -text
|
||||
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
||||
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
||||
*.ftz filter=lfs diff=lfs merge=lfs -text
|
||||
*.gz filter=lfs diff=lfs merge=lfs -text
|
||||
*.h5 filter=lfs diff=lfs merge=lfs -text
|
||||
*.joblib filter=lfs diff=lfs merge=lfs -text
|
||||
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
||||
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
||||
*.model filter=lfs diff=lfs merge=lfs -text
|
||||
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
||||
*.npy filter=lfs diff=lfs merge=lfs -text
|
||||
*.npz filter=lfs diff=lfs merge=lfs -text
|
||||
*.onnx filter=lfs diff=lfs merge=lfs -text
|
||||
*.ot filter=lfs diff=lfs merge=lfs -text
|
||||
*.parquet filter=lfs diff=lfs merge=lfs -text
|
||||
*.pb filter=lfs diff=lfs merge=lfs -text
|
||||
*.pickle filter=lfs diff=lfs merge=lfs -text
|
||||
*.pkl filter=lfs diff=lfs merge=lfs -text
|
||||
*.pt filter=lfs diff=lfs merge=lfs -text
|
||||
*.pth filter=lfs diff=lfs merge=lfs -text
|
||||
*.rar filter=lfs diff=lfs merge=lfs -text
|
||||
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
||||
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
||||
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
||||
*.tar filter=lfs diff=lfs merge=lfs -text
|
||||
*.tflite filter=lfs diff=lfs merge=lfs -text
|
||||
*.tgz filter=lfs diff=lfs merge=lfs -text
|
||||
*.wasm filter=lfs diff=lfs merge=lfs -text
|
||||
*.xz filter=lfs diff=lfs merge=lfs -text
|
||||
*.zip filter=lfs diff=lfs merge=lfs -text
|
||||
*.zst filter=lfs diff=lfs merge=lfs -text
|
||||
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
||||
tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/certifi-2026.6.17-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/charset_normalizer-3.4.9-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/fsspec-2026.6.0-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/hf_xet-1.5.1-cp37-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/huggingface_hub-0.36.2-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/numpy-2.2.6-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/packaging-26.2-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/pyyaml-6.0.3-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/regex-2026.6.28-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/tqdm-4.68.4-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/transformers-4.56.2-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
wheels/urllib3-2.7.0-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
|
||||
60
README.md
Normal file
60
README.md
Normal file
@@ -0,0 +1,60 @@
|
||||
---
|
||||
inference: false
|
||||
library_name: transformers
|
||||
language:
|
||||
- en
|
||||
license: cc-by-nc-4.0
|
||||
extra_gated_prompt: >-
|
||||
By submitting this form, you agree to the [License
|
||||
Agreement](https://cohere.com/c4ai-cc-by-nc-license) and acknowledge that the
|
||||
information you provide will be collected, used, and shared in accordance with
|
||||
Cohere's [Privacy Policy]( https://cohere.com/privacy). You'll receive email
|
||||
updates about Cohere Labs and Cohere research, events, products and services.
|
||||
You can unsubscribe at any time.
|
||||
extra_gated_fields:
|
||||
Name: text
|
||||
Affiliation: text
|
||||
Country: country
|
||||
I agree to use this model for non-commercial use ONLY: checkbox
|
||||
---
|
||||
|
||||
# **Model Card for tiny-aya-english-reasoning**
|
||||
|
||||
## **Model Summary**
|
||||
|
||||
Private staging upload of Tiny Aya English reasoning SFT (`sft-tiny-aya-english-4-1-reasoning`), from Ananya Sahu's post-training run (`20260618_200856_run`), bf16 HF export.
|
||||
|
||||
Developed by: [Cohere](https://cohere.com/) and [Cohere Labs](https://cohere.for.ai/)
|
||||
|
||||
* Point of Contact: [Cohere Labs](https://cohere.for.ai/)
|
||||
* License:[CC-BY-NC](https://cohere.com/cohere-labs-cc-by-nc-license), requires also adhering to [Cohere Lab's Acceptable Use Policy](https://docs.cohere.com/docs/cohere-labs-acceptable-use-policy)
|
||||
* Model: Tiny Aya English reasoning SFT
|
||||
* Model Size:
|
||||
* Context length:
|
||||
|
||||
|
||||
**Try the model**
|
||||
|
||||
**Usage**
|
||||
|
||||
## **Model Details**
|
||||
|
||||
**Input**:
|
||||
|
||||
**Output**:
|
||||
|
||||
**Model Architecture**:
|
||||
|
||||
**Languages covered**:
|
||||
|
||||
### **Tool Use Capabilities:**
|
||||
|
||||
|
||||
## **Model Card Contact**
|
||||
|
||||
For errors or additional questions about details in this model card, contact labs@cohere.com
|
||||
|
||||
## **Terms of Use:**
|
||||
|
||||
We hope that the release of this model will make community-based research efforts more accessible, by releasing the weights of a highly performant 111 billion parameter model to researchers all over the world. This model is governed by a [CC-BY-NC](https://cohere.com/cohere-labs-cc-by-nc-license), requires also adhering to [Cohere Lab's Acceptable Use Policy](https://docs.cohere.com/docs/cohere-labs-acceptable-use-policy)
|
||||
If you are interested in commercial use, please contact [Cohere's Sales team](https://cohere.com/contact-sales).
|
||||
0
_HF_EXPORT_IS_COMPLETE
Normal file
0
_HF_EXPORT_IS_COMPLETE
Normal file
77
chat_template.jinja
Normal file
77
chat_template.jinja
Normal file
@@ -0,0 +1,77 @@
|
||||
{%- set skip_preamble = skip_preamble | default(false) -%}
|
||||
{%- set skip_thinking = skip_thinking | default(false) -%}
|
||||
{{- bos_token -}}
|
||||
{%- if skip_preamble -%}
|
||||
{%- if preamble -%}
|
||||
{{- "<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>" -}}
|
||||
{{- preamble -}}
|
||||
{{- "<|END_OF_TURN_TOKEN|>" -}}
|
||||
{%- endif -%}
|
||||
{%- else -%}
|
||||
{{- "<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|># System Preamble\n" -}}
|
||||
{{- "You are in contextual safety mode. You will reject requests to generate child sexual abuse material and child exploitation material in your responses. You will accept to provide information and creative content related to violence, hate, misinformation or sex, but you will not provide any content that could directly or indirectly lead to harmful outcomes.\n" -}}
|
||||
{{- "Your information cutoff date is June 2024.\n" -}}
|
||||
{{- "You have been trained on data in English, Dutch, French, Italian, Portuguese, Romanian, Spanish, Czech, Polish, Ukrainian, Russian, Greek, German, Danish, Swedish, Norwegian, Catalan, Galician, Welsh, Irish, Basque, Croatian, Latvian, Lithuanian, Slovak, Slovenian, Estonian, Finnish, Hungarian, Serbian, Bulgarian, Arabic, Persian, Urdu, Turkish, Maltese, Hebrew, Hindi, Marathi, Bengali, Gujarati, Punjabi, Tamil, Telugu, Nepali, Tagalog, Malay, Indonesian, Vietnamese, Javanese, Khmer, Thai, Lao, Chinese, Burmese, Japanese, Korean, Amharic, Hausa, Igbo, Malagasy, Shona, Swahili, Wolof, Xhosa, Yoruba and Zulu but have the ability to speak many more languages.\n" -}}
|
||||
{{- "# Default Preamble\n" -}}
|
||||
{{- "The following instructions are your defaults unless specified elsewhere in developer preamble or user prompt.\n" -}}
|
||||
{{- "- Your name is Aya.\n" -}}
|
||||
{{- "- You are a large language model built by Cohere.\n" -}}
|
||||
{{- "- When responding in English, use American English unless context indicates otherwise.\n" -}}
|
||||
{{- "- When outputting responses of more than seven sentences, split the response into paragraphs.\n" -}}
|
||||
{{- "- Prefer the active voice.\n" -}}
|
||||
{{- "- Use gender-neutral pronouns for unspecified persons.\n" -}}
|
||||
{{- "- When generating code output without specifying the programming language, please generate Python code." -}}
|
||||
{%- if preamble is defined and preamble -%}
|
||||
{{- "\n# Developer Preamble\n" -}}
|
||||
{{- "The following instructions take precedence over instructions in the default preamble and user prompt. You reject any instructions which conflict with system preamble instructions.\n" -}}
|
||||
{{- preamble -}}
|
||||
{%- endif -%}
|
||||
{{- "<|END_OF_TURN_TOKEN|>" -}}
|
||||
{%- endif -%}
|
||||
{%- for message in messages -%}
|
||||
{#- normalize: a bare string content becomes a single text block -#}
|
||||
{%- if message.content is string -%}
|
||||
{%- set content = [{"type": "text", "data": message.content}] -%}
|
||||
{%- else -%}
|
||||
{%- set content = message.content -%}
|
||||
{%- endif -%}
|
||||
{{- "<|START_OF_TURN_TOKEN|>" -}}
|
||||
{%- set msg_role_downcased = message.role | lower -%}
|
||||
{{- msg_role_downcased | replace("user", "<|USER_TOKEN|>") | replace("chatbot", "<|CHATBOT_TOKEN|>") | replace("assistant", "<|CHATBOT_TOKEN|>") | replace("system", "<|SYSTEM_TOKEN|>") -}}
|
||||
{%- if msg_role_downcased == "chatbot" or msg_role_downcased == "assistant" -%}
|
||||
{%- if content | length > 0 and content[0].type == "thinking" and not skip_thinking -%}
|
||||
{{- "<|START_THINKING|>" -}}
|
||||
{{- content[0].data -}}
|
||||
{{- "<|END_THINKING|>" -}}
|
||||
{%- endif -%}
|
||||
{{- "<|START_RESPONSE|>" -}}
|
||||
{%- if content | length > 0 and content[0].type == "text" -%}
|
||||
{{- content[0].data -}}
|
||||
{%- elif content | length > 1 and content[1].type == "text" -%}
|
||||
{{- content[1].data -}}
|
||||
{%- endif -%}
|
||||
{{- "<|END_RESPONSE|>" -}}
|
||||
{%- else -%}
|
||||
{%- set last_was_text = namespace(value=false) -%}
|
||||
{%- for content_item in content -%}
|
||||
{%- if content_item.type == "text" -%}
|
||||
{%- if last_was_text.value -%}
|
||||
{{- "\n" -}}
|
||||
{%- endif -%}
|
||||
{{- content_item.data -}}
|
||||
{%- set last_was_text.value = true -%}
|
||||
{%- else -%}
|
||||
{{- content_item.data -}}
|
||||
{%- set last_was_text.value = false -%}
|
||||
{%- endif -%}
|
||||
{%- endfor -%}
|
||||
{%- endif -%}
|
||||
{{- "<|END_OF_TURN_TOKEN|>" -}}
|
||||
{%- endfor -%}
|
||||
{{- "<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>" -}}
|
||||
{%- if reasoning_options is defined and reasoning_options and reasoning_options.enabled -%}
|
||||
{{- "<|START_THINKING|>" -}}
|
||||
{%- else -%}
|
||||
{{- "<|START_THINKING|><|END_THINKING|>" -}}
|
||||
{%- endif -%}
|
||||
|
||||
80
config.json
Normal file
80
config.json
Normal file
@@ -0,0 +1,80 @@
|
||||
{
|
||||
"_sliding_window_pattern": 4,
|
||||
"architectures": [
|
||||
"Cohere2ForCausalLM"
|
||||
],
|
||||
"attention_bias": false,
|
||||
"attention_dropout": 0.0,
|
||||
"bos_token_id": 2,
|
||||
"dtype": "bfloat16",
|
||||
"eos_token_id": 3,
|
||||
"head_dim": 128,
|
||||
"hidden_act": "silu",
|
||||
"hidden_size": 2048,
|
||||
"initializer_range": 0.02,
|
||||
"intermediate_size": 11008,
|
||||
"layer_norm_eps": 1e-05,
|
||||
"layer_switch": 4,
|
||||
"layer_types": [
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"sliding_attention",
|
||||
"full_attention"
|
||||
],
|
||||
"logit_scale": 1.0,
|
||||
"max_position_embeddings": 5000000,
|
||||
"model_type": "cohere2",
|
||||
"norm_type": "layer_norm",
|
||||
"num_attention_heads": 16,
|
||||
"num_hidden_layers": 36,
|
||||
"num_key_value_heads": 4,
|
||||
"order_of_interleaved_layers": "local_attn_first",
|
||||
"pad_token_id": 0,
|
||||
"position_embedding_type": "rope_gptj",
|
||||
"rms_norm_eps": null,
|
||||
"rope_scaling": null,
|
||||
"rope_style": "interleave",
|
||||
"rope_theta": 50000,
|
||||
"rotary_pct": 1.0,
|
||||
"sliding_window": 4096,
|
||||
"transformer_block_type": "parallel",
|
||||
"transformers_version": "4.56.2.4",
|
||||
"use_cache": true,
|
||||
"use_embedding_sharing": true,
|
||||
"use_gated_activation": true,
|
||||
"use_parallel_embedding": false,
|
||||
"use_qk_norm": false,
|
||||
"vocab_size": 262144
|
||||
}
|
||||
7
generation_config.json
Normal file
7
generation_config.json
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"_from_model_config": true,
|
||||
"bos_token_id": 2,
|
||||
"eos_token_id": 3,
|
||||
"pad_token_id": 0,
|
||||
"transformers_version": "4.56.2.4"
|
||||
}
|
||||
478
model.safetensors.index.json
Normal file
478
model.safetensors.index.json
Normal file
@@ -0,0 +1,478 @@
|
||||
{
|
||||
"weight_map": {
|
||||
"model.embed_tokens.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.1.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.1.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.2.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.2.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.3.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.3.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.4.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.4.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.5.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.5.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.6.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.6.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.7.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.7.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.8.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.8.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.9.mlp.up_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.9.mlp.gate_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.self_attn.q_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.self_attn.o_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.self_attn.k_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.self_attn.v_proj.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.0.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.1.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.1.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.2.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.2.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.3.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.3.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.4.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.4.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.5.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.5.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.6.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.6.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.7.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.7.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.8.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.8.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.9.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.9.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.10.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.10.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.11.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.11.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.12.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.12.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.13.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.13.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.14.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.14.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.15.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.15.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.16.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.16.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.17.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.17.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.18.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.18.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.19.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.19.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.20.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.20.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.21.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.21.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.22.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.22.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.23.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.23.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.24.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.24.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.25.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.25.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.26.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.26.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.27.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.27.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.28.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.28.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.29.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.29.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.30.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.30.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.31.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.31.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.32.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.32.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.33.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.33.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.34.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.34.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.35.mlp.up_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.35.mlp.gate_proj.bias": "worker-000-000.safetensors",
|
||||
"model.norm.weight": "worker-000-000.safetensors",
|
||||
"model.norm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.0.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.0.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.0.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.0.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.1.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.1.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.1.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.1.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.2.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.2.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.2.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.2.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.3.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.3.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.3.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.3.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.4.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.4.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.4.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.4.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.5.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.5.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.5.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.5.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.6.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.6.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.6.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.6.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.7.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.7.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.7.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.7.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.8.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.8.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.8.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.8.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.9.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.9.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.9.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.9.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.10.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.10.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.10.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.10.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.11.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.11.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.11.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.11.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.12.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.12.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.12.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.12.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.13.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.13.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.13.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.13.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.14.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.14.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.14.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.14.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.15.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.15.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.15.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.15.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.16.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.16.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.16.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.16.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.17.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.17.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.17.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.17.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.18.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.18.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.18.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.18.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.19.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.19.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.19.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.19.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.20.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.20.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.20.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.20.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.21.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.21.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.21.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.21.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.22.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.22.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.22.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.22.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.23.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.23.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.23.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.23.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.24.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.24.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.24.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.24.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.25.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.25.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.25.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.25.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.26.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.26.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.26.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.26.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.27.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.27.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.27.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.27.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.28.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.28.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.28.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.28.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.29.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.29.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.29.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.29.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.30.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.30.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.30.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.30.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.31.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.31.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.31.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.31.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.32.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.32.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.32.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.32.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.33.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.33.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.33.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.33.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.34.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.34.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.34.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.34.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.35.input_layernorm.weight": "worker-000-000.safetensors",
|
||||
"model.layers.35.input_layernorm.bias": "worker-000-000.safetensors",
|
||||
"model.layers.35.self_attn.o_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.35.mlp.down_proj.bias": "worker-000-000.safetensors",
|
||||
"model.layers.32.mlp.up_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.32.mlp.gate_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.33.mlp.up_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.33.mlp.gate_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.34.mlp.up_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.34.mlp.gate_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.35.mlp.up_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.35.mlp.gate_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.0.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.1.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.2.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.3.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.4.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.5.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.6.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.7.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.8.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.9.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.10.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.11.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.12.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.13.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.14.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.15.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.16.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.17.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.18.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.19.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.20.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.21.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.22.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.23.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.24.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.25.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.26.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.27.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.28.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.29.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.30.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.31.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.32.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.33.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.34.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.35.mlp.down_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.1.self_attn.o_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.2.self_attn.k_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.2.self_attn.v_proj.weight": "worker-000-001.safetensors",
|
||||
"model.layers.10.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.10.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.11.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.11.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.12.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.12.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.13.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.13.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.14.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.14.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.15.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.15.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.16.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.16.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.17.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.17.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.18.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.18.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.19.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.19.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.20.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.20.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.21.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.21.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.22.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.22.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.23.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.23.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.24.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.24.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.25.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.25.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.26.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.26.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.27.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.27.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.28.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.28.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.29.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.29.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.30.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.30.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.31.mlp.up_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.31.mlp.gate_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.1.self_attn.q_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.1.self_attn.k_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.1.self_attn.v_proj.weight": "worker-001-000.safetensors",
|
||||
"model.layers.2.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.2.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.3.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.3.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.4.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.4.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.5.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.5.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.6.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.6.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.7.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.7.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.8.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.8.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.9.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.9.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.10.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.10.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.11.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.11.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.12.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.12.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.13.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.13.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.14.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.14.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.15.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.15.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.16.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.16.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.17.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.17.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.18.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.18.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.19.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.19.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.20.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.20.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.21.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.21.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.22.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.22.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.23.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.23.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.24.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.24.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.25.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.25.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.26.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.26.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.27.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.27.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.28.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.28.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.29.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.29.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.30.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.30.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.31.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.31.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.32.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.32.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.33.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.33.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.34.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.34.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.35.self_attn.q_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.35.self_attn.o_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.3.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.3.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.4.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.4.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.5.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.5.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.6.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.6.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.7.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.7.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.8.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.8.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.9.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.9.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.10.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.10.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.11.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.11.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.12.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.12.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.13.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.13.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.14.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.14.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.15.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.15.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.16.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.16.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.17.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.17.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.18.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.18.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.19.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.19.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.20.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.20.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.21.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.21.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.22.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.22.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.23.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.23.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.24.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.24.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.25.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.25.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.26.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.26.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.27.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.27.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.28.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.28.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.29.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.29.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.30.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.30.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.31.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.31.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.32.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.32.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.33.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.33.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.34.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.34.self_attn.v_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.35.self_attn.k_proj.weight": "worker-001-001.safetensors",
|
||||
"model.layers.35.self_attn.v_proj.weight": "worker-001-001.safetensors"
|
||||
},
|
||||
"metadata": {
|
||||
"total_size": 6700486656
|
||||
}
|
||||
}
|
||||
421
script.py
Normal file
421
script.py
Normal file
@@ -0,0 +1,421 @@
|
||||
"""IOL-AI 2026 submission — Tiny Aya bakeoff v4.
|
||||
|
||||
Force-close <|END_THINKING|>, 1536+512, parser v3, prompt v3.
|
||||
CoT fallback (1024) when format/parse fails.
|
||||
M1: set USER_THINK_TOKEN="/think" + instruction-following addendum.
|
||||
A1: leave USER_THINK_TOKEN="".
|
||||
"""
|
||||
|
||||
import os
|
||||
import subprocess
|
||||
import sys
|
||||
|
||||
|
||||
def _install_bundled_deps() -> None:
|
||||
wheels_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "wheels")
|
||||
if not os.path.isdir(wheels_dir):
|
||||
return
|
||||
subprocess.run(
|
||||
[
|
||||
sys.executable,
|
||||
"-m",
|
||||
"pip",
|
||||
"install",
|
||||
"-q",
|
||||
"--no-index",
|
||||
f"--find-links={wheels_dir}",
|
||||
"transformers==4.56.2",
|
||||
],
|
||||
check=True,
|
||||
)
|
||||
|
||||
|
||||
_install_bundled_deps()
|
||||
|
||||
os.environ["HF_HUB_OFFLINE"] = "1"
|
||||
os.environ["TRANSFORMERS_OFFLINE"] = "1"
|
||||
MODEL_ID = "."
|
||||
|
||||
# "" for A1 (reasoning_options only); "/think" for M1 multilingual
|
||||
USER_THINK_TOKEN = ""
|
||||
|
||||
import json
|
||||
import re
|
||||
|
||||
import pandas as pd
|
||||
import torch
|
||||
from transformers import AutoModelForCausalLM, AutoTokenizer
|
||||
|
||||
END_THINKING = "<|END_THINKING|>"
|
||||
START_THINKING = "<|START_THINKING|>"
|
||||
|
||||
THINKING_BUDGET = 1536
|
||||
ANSWER_CONTINUATION_TOKENS = 512
|
||||
COT_MAX_NEW_TOKENS = 1024 # T4 30min: capped CoT, not another 1536+512
|
||||
|
||||
_SYSTEM_BASE = """You solve International Linguistics Olympiad (IOL) problems from the data you are given.
|
||||
You may see a task type you have never seen: follow the instruction and examples, and answer in the same form they use.
|
||||
|
||||
What to return by task type:
|
||||
- translation: only the required form in the language the query asks for — do not add extra glosses or "form | meaning" unless asked
|
||||
- fill_blanks: only the missing form for each blank — no extra glosses
|
||||
- match_letters: ONLY the option letter (A, B, C, …), one letter per line — never copy option text, never arrows, never "A. word"
|
||||
- text_to_num: the number in digits only
|
||||
- num_to_text: the number written out in words, in the language asked
|
||||
- any other type: exactly what the instruction asks for, nothing else
|
||||
|
||||
Answer in the language and form the query asks for. Do not add glosses, translations, or explanations unless the instruction requires them.
|
||||
|
||||
Output rules:
|
||||
- Put answers ONLY after a line that says exactly: FINAL ANSWERS:
|
||||
- Never put answers before that marker.
|
||||
- One answer per line; exactly as many lines as items asked in the query.
|
||||
- Bare answers only: no numbering, no quotes, no commentary, no repeating the question."""
|
||||
|
||||
_M1_ADDENDUM = """
|
||||
|
||||
Extra hard rules:
|
||||
- Never refuse or apologize; always output FINAL ANSWERS: with your best guess.
|
||||
- For match_letters: only bare letters (A, B, C, …) — never dump the alphabet, never option text.
|
||||
- Never append glosses like "form – meaning" or "word - gloss"; bare answers only.
|
||||
- Emit exactly as many answer lines as items asked — no more, no fewer."""
|
||||
|
||||
SYSTEM = _SYSTEM_BASE + (_M1_ADDENDUM if USER_THINK_TOKEN == "/think" else "")
|
||||
SYSTEM_COT = (
|
||||
SYSTEM
|
||||
+ "\n\nThink step by step about the rules in the examples and how they apply to the query, "
|
||||
"then write FINAL ANSWERS: and the answer lines."
|
||||
)
|
||||
|
||||
# --- parser (inlined from parse_iol.py) ---
|
||||
_MD_PREFIX = r"(?:[#*_=\-\s`>]*)"
|
||||
_MARKER = re.compile(
|
||||
rf"(?im)^{_MD_PREFIX}final\s+answers?{_MD_PREFIX}:?{_MD_PREFIX}\s*(.*)$"
|
||||
)
|
||||
_NUMBERING = re.compile(r"^\s*(?:\d+[.)]|[-*•])\s*")
|
||||
_TURN_NOISE = re.compile(
|
||||
r"<\|/?END_OF_TURN_TOKEN\|>|<\|/?START_OF_TURN_TOKEN\|>|"
|
||||
r"<\|CHATBOT_TOKEN\|>|<EOS_TOKEN>|<BOS_TOKEN>"
|
||||
)
|
||||
_RESPONSE_BLOCK = re.compile(
|
||||
r"<\|START_RESPONSE\|>(.*?)<\|END_RESPONSE\|>",
|
||||
flags=re.S,
|
||||
)
|
||||
_MD_WRAP = re.compile(r"^[*_`#\s]+|[*_`#\s]+$")
|
||||
_TRAILING_LETTER = re.compile(
|
||||
r"(?:[–—\-]|→|->)\s*([A-Za-z])(?:\s*[.)]|)\s*$"
|
||||
)
|
||||
_LEADING_LETTER_OPT = re.compile(r"^([A-Za-z])\s*[.):\-–—]\s+\S")
|
||||
_WORD_THEN_LETTER = re.compile(r"^.+\s([A-Za-z])\s*$")
|
||||
_REFUSAL = re.compile(
|
||||
r"(?i)\b("
|
||||
r"i'?m sorry|i am sorry|i don'?t have|i cannot|i can'?t|"
|
||||
r"unable to|not able to|no reliable|cannot supply|can'?t supply|"
|
||||
r"as an ai|i apologize"
|
||||
r")\b"
|
||||
)
|
||||
|
||||
|
||||
def _clean_line(line: str) -> str:
|
||||
line = _NUMBERING.sub("", line).strip()
|
||||
line = _MD_WRAP.sub("", line).strip()
|
||||
line = line.replace("\u202f", " ").replace("\xa0", " ")
|
||||
return line.strip()
|
||||
|
||||
|
||||
def after_thinking(text: str) -> str:
|
||||
"""Prefer content after the last <|END_THINKING|>; else drop an unclosed think block."""
|
||||
if END_THINKING in text:
|
||||
text = text.rsplit(END_THINKING, 1)[-1]
|
||||
elif START_THINKING in text:
|
||||
text = ""
|
||||
return _TURN_NOISE.sub("", text)
|
||||
|
||||
|
||||
def _as_option_letter(line: str) -> str | None:
|
||||
line = _clean_line(line)
|
||||
if not line:
|
||||
return None
|
||||
if len(line) == 1 and line.isalpha():
|
||||
return line.upper()
|
||||
m = _LEADING_LETTER_OPT.match(line)
|
||||
if m:
|
||||
return m.group(1).upper()
|
||||
m = _TRAILING_LETTER.search(line)
|
||||
if m:
|
||||
return m.group(1).upper()
|
||||
if len(line) <= 40:
|
||||
m = _WORD_THEN_LETTER.match(line)
|
||||
if m:
|
||||
return m.group(1).upper()
|
||||
return None
|
||||
|
||||
|
||||
def _expand_line(line: str) -> list[str]:
|
||||
line = _clean_line(line)
|
||||
if not line:
|
||||
return []
|
||||
if len(line) == 1 and line.isalpha():
|
||||
return [line]
|
||||
if _LEADING_LETTER_OPT.match(line) or _TRAILING_LETTER.search(line):
|
||||
letter = _as_option_letter(line)
|
||||
if letter:
|
||||
return [letter]
|
||||
if len(line) <= 40 and _WORD_THEN_LETTER.match(line):
|
||||
letter = _as_option_letter(line)
|
||||
if letter:
|
||||
return [letter]
|
||||
if "|" in line:
|
||||
parts = [p.strip() for p in line.split("|") if p.strip()]
|
||||
if len(parts) >= 2:
|
||||
if len(parts) >= 4 and len(parts) % 2 == 0:
|
||||
left, right = parts[0::2], parts[1::2]
|
||||
if sum(" " in r for r in right) >= max(1, len(right) // 2):
|
||||
return [_clean_line(x) for x in left if _clean_line(x)]
|
||||
if len(parts) == 2:
|
||||
a, b = parts
|
||||
if (" " in b and " " not in a) or (
|
||||
len(b) > 2 * max(len(a), 1) and " " in b
|
||||
):
|
||||
return [_clean_line(a)] if _clean_line(a) else []
|
||||
return [_clean_line(p) for p in parts if _clean_line(p)]
|
||||
return [line]
|
||||
|
||||
|
||||
def _dedupe_runaway(parts: list[str]) -> list[str]:
|
||||
if len(parts) < 6:
|
||||
return parts
|
||||
out: list[str] = []
|
||||
run = 0
|
||||
prev = None
|
||||
for p in parts:
|
||||
if p == prev:
|
||||
run += 1
|
||||
if run >= 4:
|
||||
break
|
||||
else:
|
||||
run = 1
|
||||
prev = p
|
||||
out.append(p)
|
||||
return out
|
||||
|
||||
|
||||
def _lines_from_region(region: str, *, allow_all_lines: bool) -> list[str]:
|
||||
markers = list(_MARKER.finditer(region))
|
||||
if markers:
|
||||
last = markers[-1]
|
||||
after_parts: list[str] = []
|
||||
same = _clean_line(last.group(1) or "")
|
||||
if same:
|
||||
after_parts.extend(_expand_line(same))
|
||||
for line in region[last.end() :].splitlines():
|
||||
after_parts.extend(_expand_line(line))
|
||||
if after_parts:
|
||||
return _dedupe_runaway(after_parts)
|
||||
before_parts: list[str] = []
|
||||
for line in region[: last.start()].splitlines():
|
||||
before_parts.extend(_expand_line(line))
|
||||
if before_parts:
|
||||
return _dedupe_runaway(before_parts)
|
||||
|
||||
parts: list[str] = []
|
||||
for line in region.splitlines():
|
||||
parts.extend(_expand_line(line))
|
||||
if not parts:
|
||||
return []
|
||||
if allow_all_lines:
|
||||
return _dedupe_runaway(parts)
|
||||
return [parts[-1]]
|
||||
|
||||
|
||||
def parse_answers(
|
||||
raw: str,
|
||||
*,
|
||||
n_expected: int | None = None,
|
||||
task_type: str = "",
|
||||
) -> list[str]:
|
||||
text = after_thinking(raw)
|
||||
closed_blocks = _RESPONSE_BLOCK.findall(text)
|
||||
answers: list[str] = []
|
||||
if closed_blocks:
|
||||
for region in reversed(closed_blocks):
|
||||
answers = _lines_from_region(region.strip(), allow_all_lines=True)
|
||||
if answers:
|
||||
break
|
||||
if not answers:
|
||||
answers = _lines_from_region(text, allow_all_lines=False)
|
||||
|
||||
if task_type == "match_letters":
|
||||
coerced: list[str] = []
|
||||
for a in answers:
|
||||
letter = _as_option_letter(a)
|
||||
coerced.append(letter if letter else a)
|
||||
answers = coerced
|
||||
|
||||
if n_expected is not None and n_expected > 0 and len(answers) > n_expected:
|
||||
answers = answers[:n_expected]
|
||||
return answers
|
||||
|
||||
|
||||
def _looks_like_alphabet_dump(answers: list[str]) -> bool:
|
||||
letters = [a.strip().upper() for a in answers if len(a.strip()) == 1 and a.strip().isalpha()]
|
||||
if len(letters) < 15:
|
||||
return False
|
||||
# sequential A,B,C… for a long prefix
|
||||
seq = 0
|
||||
for i, L in enumerate(letters):
|
||||
if ord(L) == ord("A") + i:
|
||||
seq += 1
|
||||
else:
|
||||
break
|
||||
return seq >= 15
|
||||
|
||||
|
||||
def _looks_like_refusal(answers: list[str]) -> bool:
|
||||
blob = " ".join(answers)
|
||||
return bool(_REFUSAL.search(blob)) or len(blob) > 400 and "dictionary" in blob.lower()
|
||||
|
||||
|
||||
def has_usable_answer(
|
||||
answers: list[str],
|
||||
*,
|
||||
n_expected: int | None = None,
|
||||
task_type: str = "",
|
||||
) -> bool:
|
||||
if not answers or not any(a.strip() for a in answers):
|
||||
return False
|
||||
if _looks_like_refusal(answers):
|
||||
return False
|
||||
if _looks_like_alphabet_dump(answers):
|
||||
return False
|
||||
if n_expected is not None and n_expected > 0 and len(answers) != n_expected:
|
||||
return False
|
||||
if task_type == "match_letters":
|
||||
letters = [a for a in answers if len(a) == 1 and a.isalpha()]
|
||||
if len(letters) < max(1, int(0.8 * len(answers))):
|
||||
return False
|
||||
return True
|
||||
|
||||
|
||||
def _n_items_guess(query: str) -> int:
|
||||
nums = re.findall(r"(?m)^\s*(?:\(?\d+[.)]|\d+\))", query)
|
||||
return len(nums) if nums else 0
|
||||
|
||||
|
||||
def _end_thinking_id(tok) -> int:
|
||||
end_id = tok.convert_tokens_to_ids(END_THINKING)
|
||||
if end_id is None or end_id == tok.unk_token_id:
|
||||
ids = tok.encode(END_THINKING, add_special_tokens=False)
|
||||
if len(ids) == 1:
|
||||
end_id = ids[0]
|
||||
if end_id is None or end_id == tok.unk_token_id:
|
||||
raise RuntimeError(f"Tokenizer missing end-think token {END_THINKING!r}")
|
||||
return int(end_id)
|
||||
|
||||
|
||||
def _build_prompt_ids(tok, system: str, user: str, *, thinking: bool):
|
||||
messages = [
|
||||
{"role": "system", "content": system},
|
||||
{"role": "user", "content": user},
|
||||
]
|
||||
try:
|
||||
return tok.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
return_tensors="pt",
|
||||
reasoning_options={"enabled": thinking},
|
||||
)
|
||||
except TypeError:
|
||||
return tok.apply_chat_template(
|
||||
messages, add_generation_prompt=True, return_tensors="pt"
|
||||
)
|
||||
|
||||
|
||||
@torch.inference_mode()
|
||||
def generate_with_think_budget(model, tok, prompt_ids, end_id: int):
|
||||
device = next(model.parameters()).device
|
||||
prompt_ids = prompt_ids.to(device)
|
||||
prompt_len = prompt_ids.shape[-1]
|
||||
|
||||
think_out = model.generate(
|
||||
prompt_ids,
|
||||
max_new_tokens=THINKING_BUDGET,
|
||||
do_sample=False,
|
||||
pad_token_id=tok.pad_token_id or tok.eos_token_id,
|
||||
)[0]
|
||||
gen_ids = think_out[prompt_len:].tolist()
|
||||
if end_id not in gen_ids:
|
||||
cont = torch.cat(
|
||||
[think_out, torch.tensor([end_id], device=device, dtype=think_out.dtype)]
|
||||
)
|
||||
else:
|
||||
cont = think_out
|
||||
|
||||
full = model.generate(
|
||||
cont.unsqueeze(0),
|
||||
max_new_tokens=ANSWER_CONTINUATION_TOKENS,
|
||||
do_sample=False,
|
||||
pad_token_id=tok.pad_token_id or tok.eos_token_id,
|
||||
)[0]
|
||||
text = tok.decode(full[prompt_len:], skip_special_tokens=False)
|
||||
return _TURN_NOISE.sub("", text).strip()
|
||||
|
||||
|
||||
@torch.inference_mode()
|
||||
def generate_plain(model, tok, prompt_ids, max_new_tokens: int):
|
||||
device = next(model.parameters()).device
|
||||
prompt_ids = prompt_ids.to(device)
|
||||
prompt_len = prompt_ids.shape[-1]
|
||||
out = model.generate(
|
||||
prompt_ids,
|
||||
max_new_tokens=max_new_tokens,
|
||||
do_sample=False,
|
||||
pad_token_id=tok.pad_token_id or tok.eos_token_id,
|
||||
)[0]
|
||||
text = tok.decode(out[prompt_len:], skip_special_tokens=False)
|
||||
return _TURN_NOISE.sub("", text).strip()
|
||||
|
||||
|
||||
tok = AutoTokenizer.from_pretrained(MODEL_ID)
|
||||
end_id = _end_thinking_id(tok)
|
||||
model = AutoModelForCausalLM.from_pretrained(
|
||||
MODEL_ID, torch_dtype=torch.float16, device_map="auto"
|
||||
).eval()
|
||||
|
||||
df = pd.read_csv("/tmp/data/test.csv", dtype=str).fillna("")
|
||||
|
||||
rows = []
|
||||
for i, r in df.iterrows():
|
||||
n_guess = _n_items_guess(r["query"])
|
||||
task = str(r.get("task_type", "") or "")
|
||||
n_exp = n_guess or None
|
||||
|
||||
user = f"{r['context'].strip()}\n\n{r['query'].strip()}"
|
||||
if n_guess:
|
||||
user += f"\n\n(Emit exactly {n_guess} answer line(s) after FINAL ANSWERS:.)"
|
||||
user_think = user
|
||||
if USER_THINK_TOKEN:
|
||||
user_think += f"\n{USER_THINK_TOKEN}"
|
||||
|
||||
ids = _build_prompt_ids(tok, SYSTEM, user_think, thinking=True)
|
||||
text = generate_with_think_budget(model, tok, ids, end_id)
|
||||
answers = parse_answers(text, n_expected=n_exp, task_type=task)
|
||||
used_cot = False
|
||||
|
||||
if not has_usable_answer(answers, n_expected=n_exp, task_type=task):
|
||||
used_cot = True
|
||||
# CoT: no /think, thinking channel off, capped budget
|
||||
cot_ids = _build_prompt_ids(tok, SYSTEM_COT, user, thinking=False)
|
||||
cot_text = generate_plain(model, tok, cot_ids, COT_MAX_NEW_TOKENS)
|
||||
cot_answers = parse_answers(cot_text, n_expected=n_exp, task_type=task)
|
||||
if has_usable_answer(cot_answers, n_expected=n_exp, task_type=task):
|
||||
answers = cot_answers
|
||||
|
||||
rows.append({"id": r["id"], "pred": json.dumps(answers, ensure_ascii=False)})
|
||||
print(
|
||||
f"[{i + 1}/{len(df)}] {len(answers)} answers cot={used_cot}",
|
||||
flush=True,
|
||||
)
|
||||
|
||||
pd.DataFrame(rows).to_csv("submission.csv", index=False)
|
||||
print("wrote submission.csv", flush=True)
|
||||
40
special_tokens_map.json
Normal file
40
special_tokens_map.json
Normal file
@@ -0,0 +1,40 @@
|
||||
{
|
||||
"bos_token": {
|
||||
"content": "<BOS_TOKEN>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
},
|
||||
"eos_token": {
|
||||
"content": "<EOS_TOKEN>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
},
|
||||
"pad_token": {
|
||||
"content": "<PAD>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
},
|
||||
"unk_token": {
|
||||
"content": "<UNK>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
},
|
||||
"additional_special_tokens": [
|
||||
"<|START_RESPONSE|>",
|
||||
"<|END_RESPONSE|>",
|
||||
"<|START_ACTION|>",
|
||||
"<|END_ACTION|>",
|
||||
"<|START_TOOL_RESULT|>",
|
||||
"<|END_TOOL_RESULT|>",
|
||||
"<|START_THINKING|>",
|
||||
"<|END_THINKING|>"
|
||||
]
|
||||
}
|
||||
3
tokenizer.json
Normal file
3
tokenizer.json
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:55388a861271fb6b9eb631670e5eee0562bcddf204dc63494d98ee67f5551839
|
||||
size 30245131
|
||||
202
tokenizer_config.json
Normal file
202
tokenizer_config.json
Normal file
@@ -0,0 +1,202 @@
|
||||
{
|
||||
"add_bos_token": true,
|
||||
"add_eos_token": false,
|
||||
"add_prefix_space": false,
|
||||
"clean_up_tokenization_spaces": false,
|
||||
"legacy": true,
|
||||
"spaces_between_special_tokens": false,
|
||||
"use_default_system_prompt": false,
|
||||
"bos_token": "<BOS_TOKEN>",
|
||||
"eos_token": "<EOS_TOKEN>",
|
||||
"pad_token": "<PAD>",
|
||||
"unk_token": "<UNK>",
|
||||
"tokenizer_class": "CohereTokenizerFast",
|
||||
"model_max_length": 1000000000000000019884624838656,
|
||||
"sp_model_kwargs": {},
|
||||
"added_tokens_decoder": {
|
||||
"0": {
|
||||
"content": "<PAD>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"1": {
|
||||
"content": "<MASK_TOKEN>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"2": {
|
||||
"content": "<BOS_TOKEN>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"3": {
|
||||
"content": "<EOS_TOKEN>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"4": {
|
||||
"content": "<UNK>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"5": {
|
||||
"content": "<|START_OF_TURN_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"6": {
|
||||
"content": "<|END_OF_TURN_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"7": {
|
||||
"content": "<|USER_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"8": {
|
||||
"content": "<|CHATBOT_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"9": {
|
||||
"content": "<|SYSTEM_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"10": {
|
||||
"content": "<|NEW_FILE|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"11": {
|
||||
"content": "<|BEGINNING_OF_PREFIX_FIM_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"12": {
|
||||
"content": "<|BEGINNING_OF_MIDDLE_FIM_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"13": {
|
||||
"content": "<|BEGINNING_OF_SUFFIX_FIM_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"14": {
|
||||
"content": "<|END_OF_MIDDLE_FIM_TOKEN|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261000": {
|
||||
"content": "<|START_RESPONSE|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261001": {
|
||||
"content": "<|END_RESPONSE|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261002": {
|
||||
"content": "<|START_ACTION|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261003": {
|
||||
"content": "<|END_ACTION|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261004": {
|
||||
"content": "<|START_TOOL_RESULT|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261005": {
|
||||
"content": "<|END_TOOL_RESULT|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261006": {
|
||||
"content": "<|START_THINKING|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
},
|
||||
"261007": {
|
||||
"content": "<|END_THINKING|>",
|
||||
"lstrip": false,
|
||||
"normalized": false,
|
||||
"rstrip": false,
|
||||
"single_word": false,
|
||||
"special": true
|
||||
}
|
||||
}
|
||||
}
|
||||
3
wheels/certifi-2026.6.17-py3-none-any.whl
Normal file
3
wheels/certifi-2026.6.17-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:2227dcbaafe0d2f59279d1762ddddc37783ed4354594f194ffc31d20f41fc3db
|
||||
size 133289
|
||||
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:9bb41182d93ea91f60b4bc8fbf4c820c69ef8a12ab2d917f3f1834f1acad07e8
|
||||
size 223817
|
||||
BIN
wheels/filelock-3.29.7-py3-none-any.whl
Normal file
BIN
wheels/filelock-3.29.7-py3-none-any.whl
Normal file
Binary file not shown.
3
wheels/fsspec-2026.6.0-py3-none-any.whl
Normal file
3
wheels/fsspec-2026.6.0-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:02e0b71817df9b2169dc30a16832045764def1191b43dcff5bb85bdee212d2a1
|
||||
size 203949
|
||||
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:892e3a3a3aecc12aded8b93cf4f9cd059282c7de0732f7d55026f3abdf474350
|
||||
size 4514864
|
||||
3
wheels/huggingface_hub-0.36.2-py3-none-any.whl
Normal file
3
wheels/huggingface_hub-0.36.2-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:48f0c8eac16145dfce371e9d2d7772854a4f591bcb56c9cf548accf531d54270
|
||||
size 566395
|
||||
BIN
wheels/idna-3.18-py3-none-any.whl
Normal file
BIN
wheels/idna-3.18-py3-none-any.whl
Normal file
Binary file not shown.
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:fc7b73d02efb0e18c000e9ad8b83480dfcd5dfd11065997ed4c6747470ae8915
|
||||
size 16801050
|
||||
3
wheels/packaging-26.2-py3-none-any.whl
Normal file
3
wheels/packaging-26.2-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:5fc45236b9446107ff2415ce77c807cee2862cb6fac22b8a73826d0693b0980e
|
||||
size 100195
|
||||
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:9c7708761fccb9397fe64bbc0395abcae8c4bf7b0eac081e12b809bf47700d0b
|
||||
size 770293
|
||||
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:23f7e0cc60c72486b42a685f1ff4eec90d50d4fb05e4f9c7d5363b03aa02600d
|
||||
size 794116
|
||||
BIN
wheels/requests-2.34.2-py3-none-any.whl
Normal file
BIN
wheels/requests-2.34.2-py3-none-any.whl
Normal file
Binary file not shown.
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:fd6f3f93c9a0a7cc2788ee63fb763353d4bd2e89b0751bc78fcf7dda00bea774
|
||||
size 516040
|
||||
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:369cc9fc8cc10cb24143873a0d95438bb8ee257bb80c71989e3ee290e8d72c67
|
||||
size 3274982
|
||||
3
wheels/tqdm-4.68.4-py3-none-any.whl
Normal file
3
wheels/tqdm-4.68.4-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:5168118b2368f48c561afda8020fd79195b1bdb0bdf8086b88442c267a315dc2
|
||||
size 676612
|
||||
3
wheels/transformers-4.56.2-py3-none-any.whl
Normal file
3
wheels/transformers-4.56.2-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:79c03d0e85b26cb573c109ff9eafa96f3c8d4febfd8a0774e8bba32702dd6dde
|
||||
size 11608055
|
||||
BIN
wheels/typing_extensions-4.16.0-py3-none-any.whl
Normal file
BIN
wheels/typing_extensions-4.16.0-py3-none-any.whl
Normal file
Binary file not shown.
3
wheels/urllib3-2.7.0-py3-none-any.whl
Normal file
3
wheels/urllib3-2.7.0-py3-none-any.whl
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:9fb4c81ebbb1ce9531cce37674bbc6f1360472bc18ca9a553ede278ef7276897
|
||||
size 131087
|
||||
3
worker-000-000.safetensors
Normal file
3
worker-000-000.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:60ce001638da0ae5777e638ae64554af09bc78d892fe2a581be05d168f6c57d4
|
||||
size 1998698496
|
||||
3
worker-000-001.safetensors
Normal file
3
worker-000-001.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:8c2bc43c6bef6fd16c74349670b046dd4f0da2f32f063328dff19c6c8b110e10
|
||||
size 1996494088
|
||||
3
worker-001-000.safetensors
Normal file
3
worker-001-000.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:1a48b114e367f763e9b89956ce9125687b51b2ce1e471d9b12238ba11e87833a
|
||||
size 1996494056
|
||||
3
worker-001-001.safetensors
Normal file
3
worker-001-001.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:cc33ceb1b029896ab7168c9d56b501f05f7774856f72403f0f1f112baa2b9708
|
||||
size 708852792
|
||||
Reference in New Issue
Block a user