diff --git a/EVAL_REPORT.md b/EVAL_REPORT.md
new file mode 100644
index 0000000000000000000000000000000000000000..6e63ad782ec3ef8321668a03daad66db0a6581e5
--- /dev/null
+++ b/EVAL_REPORT.md
@@ -0,0 +1,144 @@
+# Automated MNLP evaluation report
+
+- **Model repo:** [`cs-552-2026-the-transformers/math_model`](https://huggingface.co/cs-552-2026-the-transformers/math_model)
+- **Owner(s):** group **the-transformers**
+- **Generated at:** 2026-06-11T06:23:10+00:00 (UTC)
+- **Pipeline:** [mnlp-project-ci](https://github.com/eric11eca/mnlp-project-ci)
+
+_This PR is opened automatically by the course CI. It is **non-blocking** — you do not need to merge it. The next nightly run will refresh this file._
+
+## Evaluated checkpoint
+
+- **Commit:** [`f9658b7`](https://huggingface.co/cs-552-2026-the-transformers/math_model/commit/f9658b7782b5e50dd8aa2a81f67c1d03d3476f92)
+- **Message:** Update model
+- **Committed:** 2026-06-10T17:25:31+00:00
+
+## Summary
+
+| Benchmark | Accuracy | Status |
+|---|---:|---|
+| Math | 0.5300 | ok |
+| Knowledge | — | not run |
+| Multilingual | — | not run |
+| Safety | — | not run |
+
+## Sample completions
+
+_Prompts are intentionally omitted to avoid revealing benchmark contents. For multi-completion problems, only one completion is shown per sample._
+
+### Math
+
+**Correct** (1 shown)
+
+- **reference**: `40_9`
+- **overall** (7/8 completions correct)
+- **extracted** (✓): `40_9`
+- **completion**:
+
+ ```text
+
+ Okay, let's see. I need to subtract two numbers in base 9: 58₉ minus 18₉. Hmm, base 9... I remember that subtracting in different bases can be similar to decimal, but you have to remember that each digit goes from 0 to 8 instead of 0 to 9. So maybe I should convert both numbers to base 10, subtract them, and then convert the result back to base 9? That might be easier for me, since I'm mor
+
+ ... [truncated 823 chars] ...
+
+ I think the answer is correct.
+
+
+ To solve the problem of evaluating \(58_9 - 18_9\) and writing the result in base 9, we can follow these steps:
+
+ 1. **Convert both numbers to base 10:**
+ - Convert \(58_9\) to base 10:
+ \[
+ 58_9 = 5 \times 9^1 + 8 \times 9^0 = 5 \times 9 + 8 = 45 + 8 = 53_{10}
+ \]
+ - Convert \(18_9\) to base 10:
+ \[
+ 18_9 = 1 \times 9^1 + 8 \times 9^0 = 1 \times 9 + 8 = 9 + 8 = 17_{10}
+ \]
+
+ 2. **Subtract the two base 10 numbers:**
+ \[
+ 53_{10} - 17_{10} = 36_{10}
+ \]
+
+ 3. **Convert the result back to base 9:**
+ - Divide 36 by 9 to find the quotient and remainder:
+ \[
+ 36 \div 9 = 4 \text{ remainder } 0
+ \]
+ - Therefore, \(36_{10}\) in base 9 is \(40_9\).
+
+ 4. **Verify the result:**
+ - Convert \(40_9\) back to base 10:
+ \[
+ 40_9 = 4 \times 9^1 + 0 \times 9^0 = 4 \times 9 + 0 = 36_{10}
+ \]
+ - Convert \(58_9\) and \(18_9\) back to base 10:
+ \[
+ 58_9 = 53_{10}, \quad 18_9 = 17_{10}
+ \]
+ - Subtract the base 10 numbers:
+ \[
+ 53_{10} - 17_{10} = 36_{10}
+ \]
+ - Convert \(36_{10}\) back to base 9:
+ \[
+ 36_{10} = 40_9
+ \]
+
+ Thus, the final answer is \(\boxed{40_9}\).
+ ```
+
+**Incorrect** (1 shown)
+
+- **reference**: `$f(x,y)= g(x+y, xy(x-y)^{2})$ for some polynomial $g$`
+- **overall** (0/8 completions correct)
+- **extracted** (✗): `c`
+- **completion**:
+
+ ```text
+
+ Okay, let's try to figure out this problem. So, we need to find all polynomials f(x, y) in complex numbers such that for any complex numbers a and b, the equation f(a², b²) equals f((a - b)² / 2, (a + b)² / 2). Hmm, interesting. Let me start by understanding what this condition means.
+
+ First, maybe I should consider specific values for a and b to see if I can spot a pattern or figure out t
+
+ ... [truncated 822 chars] ...
+
+ + q}{2}\right)a^2 + \left(\frac{p + q}{2}\right)b^2 - pqab + r.
+ \]
+ Equating the coefficients with the left-hand side:
+ \[
+ p a^2 + q b^2 + r = \left(\frac{p + q}{2}\right)a^2 + \left(\frac{p + q}{2}\right)b^2 - pqab + r.
+ \]
+ For this to hold for all $ a $ and $ b $, the coefficients must match:
+ \[
+ p = \frac{p + q}{2} \implies 2p = p + q \implies q = p,
+ \]
+ \[
+ q = \frac{p + q}{2} \implies 2q = p + q \implies p = q,
+ \]
+ \[
+ -pq = 0 \implies pq = 0 \implies p = 0 \text{ (since } p \neq 0).
+ \]
+ Thus, the only solution for linear polynomials is $ f(x, y) = 0 $, which is a constant.
+
+ #### Quadratic Polynomials
+
+ Assume $ f(x, y) = Ax^2 + By^2 + Cxy + Dx + Ey + F $. Substitute into the equation and equate coefficients. This leads to:
+ \[
+ A = B = C = D = E = F = 0.
+ \]
+ Thus, the only solution is $ f(x, y) = c $.
+
+ ### Step 4: Higher-Degree Polynomials
+
+ Higher-degree polynomials will similarly force all coefficients to zero, confirming that only constant polynomials satisfy the condition.
+
+ ### Conclusion
+
+ After testing various forms and deriving constraints, the only polynomials $ f(x, y) \in \mathbb{C}[x, y] \) that satisfy the given functional equation are the constant polynomials.
+
+ $$
+ \boxed{c}
+ $$
+ ```
diff --git a/README.md b/README.md
index 863e7d7f17ef0110a0719c682b315ea01f4e81a4..88ac3cf98ab04cb3a6c98e1025b34dcd85b918a6 100644
--- a/README.md
+++ b/README.md
@@ -2,56 +2,10 @@
base_model: Qwen/Qwen3-1.7B
library_name: transformers
tags:
-- generated_from_trainer
-- trl
+- math
- sft
-licence: license
---
# math_model
-This model is a fine-tuned version of [Qwen/Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B).
-It has been trained using [TRL](https://github.com/huggingface/trl).
-
-## Quick start
-
-```python
-from transformers import pipeline
-
-question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
-generator = pipeline("text-generation", model="cs-552-2026-the-transformers/math_model", device="cuda")
-output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
-print(output["generated_text"])
-```
-
-## Training procedure
-
-[
](https://wandb.ai/jdecim/MNLP/runs/2btc84l3)
-
-
-This model was trained with SFT.
-
-### Framework versions
-
-- TRL: 0.27.2
-- Transformers: 5.8.0
-- Pytorch: 2.10.0+cu128
-- Datasets: 4.8.5
-- Tokenizers: 0.22.2
-
-## Citations
-
-
-
-Cite TRL as:
-
-```bibtex
-@misc{vonwerra2022trl,
- title = {{TRL: Transformer Reinforcement Learning}},
- author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
- year = 2020,
- journal = {GitHub repository},
- publisher = {GitHub},
- howpublished = {\url{https://github.com/huggingface/trl}}
-}
-```
\ No newline at end of file
+Fine-tuned version of [Qwen/Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B) for math reasoning.
diff --git a/chat_template.jinja b/chat_template.jinja
index 5845311e60103eb4a734e3c4c8e69ba7348636a7..01be9b307daa2d425f7c168c9fb145a286e0afb4 100644
--- a/chat_template.jinja
+++ b/chat_template.jinja
@@ -10,13 +10,9 @@
{%- endfor %}
{{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
{%- else %}
- {#- BOXING INSTRUCTION injected so it is present even when the caller (e.g. CI)
- passes no system message. Targets our no-box / multi-box failure mode. #}
- {{- '<|im_start|>system\n' }}
{%- if messages[0].role == 'system' %}
- {{- messages[0].content + '\n\n' }}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
{%- endif %}
- {{- 'When solving a math problem, work through it step by step and write your final answer inside a \boxed{...}.<|im_end|>\n' }}
{%- endif %}
{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
{%- for message in messages[::-1] %}
diff --git a/checkpoint-1011/chat_template.jinja b/checkpoint-1011/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..01be9b307daa2d425f7c168c9fb145a286e0afb4
--- /dev/null
+++ b/checkpoint-1011/chat_template.jinja
@@ -0,0 +1,89 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0].role == 'system' %}
+ {{- messages[0].content + '\n\n' }}
+ {%- endif %}
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0].role == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+ {%- set index = (messages|length - 1) - loop.index0 %}
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}
+ {%- set ns.multi_step_tool = false %}
+ {%- set ns.last_query_index = index %}
+ {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+ {%- if message.content is string %}
+ {%- set content = message.content %}
+ {%- else %}
+ {%- set content = '' %}
+ {%- endif %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {%- set reasoning_content = '' %}
+ {%- if message.reasoning_content is string %}
+ {%- set reasoning_content = message.reasoning_content %}
+ {%- else %}
+ {%- if '' in content %}
+ {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %}
+ {%- set content = content.split('')[-1].lstrip('\n') %}
+ {%- endif %}
+ {%- endif %}
+ {%- if loop.index0 > ns.last_query_index %}
+ {%- if loop.last or (not loop.last and reasoning_content) %}
+ {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- if message.tool_calls %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if (loop.first and content) or (not loop.first) %}
+ {{- '\n' }}
+ {%- endif %}
+ {%- if tool_call.function %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {%- if tool_call.arguments is string %}
+ {{- tool_call.arguments }}
+ {%- else %}
+ {{- tool_call.arguments | tojson }}
+ {%- endif %}
+ {{- '}\n' }}
+ {%- endfor %}
+ {%- endif %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+ {%- if enable_thinking is defined and enable_thinking is false %}
+ {{- '\n\n\n\n' }}
+ {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/checkpoint-1011/config.json b/checkpoint-1011/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d219d6a6d288625229a71646710b85cf7ae5def
--- /dev/null
+++ b/checkpoint-1011/config.json
@@ -0,0 +1,63 @@
+{
+ "architectures": [
+ "Qwen3ForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "bfloat16",
+ "eos_token_id": 151645,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 6144,
+ "layer_types": [
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 40960,
+ "max_window_layers": 28,
+ "model_type": "qwen3",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 8,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.7.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/checkpoint-1011/generation_config.json b/checkpoint-1011/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec133d867b1427d43a5ee0ea332bd088fc12013
--- /dev/null
+++ b/checkpoint-1011/generation_config.json
@@ -0,0 +1,12 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "temperature": 0.6,
+ "top_k": 20,
+ "top_p": 0.95,
+ "transformers_version": "5.7.0"
+}
diff --git a/checkpoint-1011/model.safetensors b/checkpoint-1011/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cc885a501a0fffef97b843b5248e3bb7cfe46172
--- /dev/null
+++ b/checkpoint-1011/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3af4557502ae7c7853ed3b684aa50cfd00317aafc237f2dda688075bb3dc81c1
+size 3441185608
diff --git a/checkpoint-1011/optimizer.pt b/checkpoint-1011/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d00b4ce7da17dce3a0fb31d9c369e0cafbfc68a3
--- /dev/null
+++ b/checkpoint-1011/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:016daefe3fcafd2822ed799155138cb8e476fe309f9070f5157970a18bcc6d4c
+size 6882572207
diff --git a/checkpoint-1011/rng_state.pth b/checkpoint-1011/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..bd3c190a62ba12f21428f29e0f4bde711034a75b
--- /dev/null
+++ b/checkpoint-1011/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4a9f217e852f439efa6bd32fde98d6867f11aa6ea13ddc021ba10af6a0b0934
+size 14645
diff --git a/checkpoint-1011/scheduler.pt b/checkpoint-1011/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..edf0d85a22eaf7a475ed68221fba576eaa55f682
--- /dev/null
+++ b/checkpoint-1011/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ee946d348a6273229e02f92d4563cb2ba7ac75075130a9d6c5c0eb31d584ddc
+size 1465
diff --git a/checkpoint-1011/tokenizer.json b/checkpoint-1011/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7afbed2efcdf019f88ab0572ec29d3bf595dfe2
--- /dev/null
+++ b/checkpoint-1011/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650
diff --git a/checkpoint-1011/tokenizer_config.json b/checkpoint-1011/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/checkpoint-1011/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "local_files_only": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/checkpoint-1011/trainer_state.json b/checkpoint-1011/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9becb660138c318878d26d4efdaa5944723c7ae4
--- /dev/null
+++ b/checkpoint-1011/trainer_state.json
@@ -0,0 +1,1044 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 500,
+ "global_step": 1011,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "entropy": NaN,
+ "epoch": 0.029717682020802376,
+ "grad_norm": NaN,
+ "learning_rate": 1.8000000000000001e-06,
+ "loss": 1005514863411.2,
+ "mean_token_accuracy": 0.07770282882993343,
+ "num_tokens": 97764.0,
+ "step": 10
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.05943536404160475,
+ "grad_norm": NaN,
+ "learning_rate": 3.8000000000000005e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007392044077278115,
+ "num_tokens": 190372.0,
+ "step": 20
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.08915304606240713,
+ "grad_norm": NaN,
+ "learning_rate": 5.8e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000383764869911829,
+ "num_tokens": 280928.0,
+ "step": 30
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.1188707280832095,
+ "grad_norm": NaN,
+ "learning_rate": 7.800000000000002e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003212536917999387,
+ "num_tokens": 380644.0,
+ "step": 40
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.1485884101040119,
+ "grad_norm": NaN,
+ "learning_rate": 9.800000000000001e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003853117574180942,
+ "num_tokens": 475467.0,
+ "step": 50
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.17830609212481427,
+ "grad_norm": NaN,
+ "learning_rate": 1.18e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003134153150313068,
+ "num_tokens": 572411.0,
+ "step": 60
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.20802377414561665,
+ "grad_norm": NaN,
+ "learning_rate": 1.38e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011001345737895463,
+ "num_tokens": 661517.0,
+ "step": 70
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.237741456166419,
+ "grad_norm": NaN,
+ "learning_rate": 1.58e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012231820757733658,
+ "num_tokens": 752142.0,
+ "step": 80
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.2674591381872214,
+ "grad_norm": NaN,
+ "learning_rate": 1.7800000000000002e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011551132018212229,
+ "num_tokens": 850399.0,
+ "step": 90
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.2971768202080238,
+ "grad_norm": NaN,
+ "learning_rate": 1.98e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00047943509634933433,
+ "num_tokens": 954759.0,
+ "step": 100
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.32689450222882616,
+ "grad_norm": NaN,
+ "learning_rate": 1.9802414928649837e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008438902725174558,
+ "num_tokens": 1050439.0,
+ "step": 110
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.35661218424962854,
+ "grad_norm": NaN,
+ "learning_rate": 1.958287596048299e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009903057703922967,
+ "num_tokens": 1135089.0,
+ "step": 120
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.3863298662704309,
+ "grad_norm": NaN,
+ "learning_rate": 1.9363336992316137e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00040032515971688554,
+ "num_tokens": 1228149.0,
+ "step": 130
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.4160475482912333,
+ "grad_norm": NaN,
+ "learning_rate": 1.9143798024149288e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0010247966441966128,
+ "num_tokens": 1331672.0,
+ "step": 140
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.4457652303120357,
+ "grad_norm": NaN,
+ "learning_rate": 1.892425905598244e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000843161555530969,
+ "num_tokens": 1421314.0,
+ "step": 150
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.475482912332838,
+ "grad_norm": NaN,
+ "learning_rate": 1.8704720087815587e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007889680869993754,
+ "num_tokens": 1511701.0,
+ "step": 160
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5052005943536404,
+ "grad_norm": NaN,
+ "learning_rate": 1.848518111964874e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007493881763366517,
+ "num_tokens": 1605063.0,
+ "step": 170
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5349182763744428,
+ "grad_norm": NaN,
+ "learning_rate": 1.826564215148189e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000567350817436818,
+ "num_tokens": 1703107.0,
+ "step": 180
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5646359583952452,
+ "grad_norm": NaN,
+ "learning_rate": 1.804610318331504e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00027060438151238486,
+ "num_tokens": 1796279.0,
+ "step": 190
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5943536404160475,
+ "grad_norm": NaN,
+ "learning_rate": 1.782656421514819e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008226798396208324,
+ "num_tokens": 1893647.0,
+ "step": 200
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.6240713224368499,
+ "grad_norm": NaN,
+ "learning_rate": 1.760702524698134e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007338574505411088,
+ "num_tokens": 1982325.0,
+ "step": 210
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.6537890044576523,
+ "grad_norm": NaN,
+ "learning_rate": 1.7387486278814492e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004891513846814633,
+ "num_tokens": 2071869.0,
+ "step": 220
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.6835066864784547,
+ "grad_norm": NaN,
+ "learning_rate": 1.716794731064764e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00038509477453771976,
+ "num_tokens": 2168370.0,
+ "step": 230
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.7132243684992571,
+ "grad_norm": NaN,
+ "learning_rate": 1.694840834248079e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009255842014681548,
+ "num_tokens": 2263211.0,
+ "step": 240
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.7429420505200595,
+ "grad_norm": NaN,
+ "learning_rate": 1.6728869374313943e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011857455348945223,
+ "num_tokens": 2352425.0,
+ "step": 250
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.7726597325408618,
+ "grad_norm": NaN,
+ "learning_rate": 1.6509330406147094e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008204675337765366,
+ "num_tokens": 2445217.0,
+ "step": 260
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.8023774145616642,
+ "grad_norm": NaN,
+ "learning_rate": 1.6289791437980242e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0013553448639868292,
+ "num_tokens": 2537677.0,
+ "step": 270
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.8320950965824666,
+ "grad_norm": NaN,
+ "learning_rate": 1.6070252469813393e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011630529144895263,
+ "num_tokens": 2631646.0,
+ "step": 280
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.861812778603269,
+ "grad_norm": NaN,
+ "learning_rate": 1.5850713501646545e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009409735575900413,
+ "num_tokens": 2722632.0,
+ "step": 290
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.8915304606240714,
+ "grad_norm": NaN,
+ "learning_rate": 1.5631174533479693e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006484775040007662,
+ "num_tokens": 2819185.0,
+ "step": 300
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.9212481426448736,
+ "grad_norm": NaN,
+ "learning_rate": 1.5411635565312844e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011682231881422922,
+ "num_tokens": 2910483.0,
+ "step": 310
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.950965824665676,
+ "grad_norm": NaN,
+ "learning_rate": 1.5192096597145995e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004708930806373246,
+ "num_tokens": 2997057.0,
+ "step": 320
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.9806835066864784,
+ "grad_norm": NaN,
+ "learning_rate": 1.4972557628979145e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011955970534472727,
+ "num_tokens": 3091284.0,
+ "step": 330
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.0089153046062407,
+ "grad_norm": NaN,
+ "learning_rate": 1.4753018660812295e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006843029964090276,
+ "num_tokens": 3177800.0,
+ "step": 340
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.0386329866270432,
+ "grad_norm": NaN,
+ "learning_rate": 1.4533479692645444e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0005406756106822285,
+ "num_tokens": 3271989.0,
+ "step": 350
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.0683506686478454,
+ "grad_norm": NaN,
+ "learning_rate": 1.4313940724478596e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007009803419350646,
+ "num_tokens": 3365082.0,
+ "step": 360
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.098068350668648,
+ "grad_norm": NaN,
+ "learning_rate": 1.4094401756311746e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001003894711175235,
+ "num_tokens": 3465865.0,
+ "step": 370
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.1277860326894502,
+ "grad_norm": NaN,
+ "learning_rate": 1.3874862788144897e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004136150790145621,
+ "num_tokens": 3564998.0,
+ "step": 380
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.1575037147102525,
+ "grad_norm": NaN,
+ "learning_rate": 1.3655323819978048e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000998112405068241,
+ "num_tokens": 3661508.0,
+ "step": 390
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.187221396731055,
+ "grad_norm": NaN,
+ "learning_rate": 1.3435784851811198e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009621107514249161,
+ "num_tokens": 3755395.0,
+ "step": 400
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.2169390787518575,
+ "grad_norm": NaN,
+ "learning_rate": 1.3216245883644348e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004976072828867473,
+ "num_tokens": 3842994.0,
+ "step": 410
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.2466567607726597,
+ "grad_norm": NaN,
+ "learning_rate": 1.2996706915477499e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003434549740632065,
+ "num_tokens": 3934618.0,
+ "step": 420
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.276374442793462,
+ "grad_norm": NaN,
+ "learning_rate": 1.2777167947310649e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00026364591685705817,
+ "num_tokens": 4026711.0,
+ "step": 430
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3060921248142645,
+ "grad_norm": NaN,
+ "learning_rate": 1.2557628979143798e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006349188319290988,
+ "num_tokens": 4114609.0,
+ "step": 440
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3358098068350668,
+ "grad_norm": NaN,
+ "learning_rate": 1.2338090010976948e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004265022376785055,
+ "num_tokens": 4203953.0,
+ "step": 450
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3655274888558693,
+ "grad_norm": NaN,
+ "learning_rate": 1.2118551042810101e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012633154692593962,
+ "num_tokens": 4298319.0,
+ "step": 460
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3952451708766715,
+ "grad_norm": NaN,
+ "learning_rate": 1.189901207464325e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011256658195634372,
+ "num_tokens": 4396020.0,
+ "step": 470
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.424962852897474,
+ "grad_norm": NaN,
+ "learning_rate": 1.16794731064764e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0013131745137798135,
+ "num_tokens": 4491203.0,
+ "step": 480
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.4546805349182763,
+ "grad_norm": NaN,
+ "learning_rate": 1.1459934138309552e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012716218378045596,
+ "num_tokens": 4582199.0,
+ "step": 490
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.4843982169390788,
+ "grad_norm": NaN,
+ "learning_rate": 1.1240395170142701e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0014869970778818241,
+ "num_tokens": 4672589.0,
+ "step": 500
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.514115898959881,
+ "grad_norm": NaN,
+ "learning_rate": 1.1020856201975851e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00046981508567114363,
+ "num_tokens": 4767693.0,
+ "step": 510
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.5438335809806834,
+ "grad_norm": NaN,
+ "learning_rate": 1.0801317233809e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000601445968641201,
+ "num_tokens": 4860371.0,
+ "step": 520
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.5735512630014858,
+ "grad_norm": NaN,
+ "learning_rate": 1.0581778265642154e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012073652920662425,
+ "num_tokens": 4950622.0,
+ "step": 530
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6032689450222883,
+ "grad_norm": NaN,
+ "learning_rate": 1.0362239297475303e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00028707647215924225,
+ "num_tokens": 5044878.0,
+ "step": 540
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6329866270430906,
+ "grad_norm": NaN,
+ "learning_rate": 1.0142700329308453e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004101752907445189,
+ "num_tokens": 5140938.0,
+ "step": 550
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6627043090638929,
+ "grad_norm": NaN,
+ "learning_rate": 9.923161361141603e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008072987395280507,
+ "num_tokens": 5232852.0,
+ "step": 560
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6924219910846954,
+ "grad_norm": NaN,
+ "learning_rate": 9.703622392974754e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012222853780258446,
+ "num_tokens": 5321835.0,
+ "step": 570
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.7221396731054979,
+ "grad_norm": NaN,
+ "learning_rate": 9.484083424807904e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001064640874392353,
+ "num_tokens": 5416464.0,
+ "step": 580
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.7518573551263001,
+ "grad_norm": NaN,
+ "learning_rate": 9.264544456641055e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009612867230316624,
+ "num_tokens": 5515169.0,
+ "step": 590
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.7815750371471024,
+ "grad_norm": NaN,
+ "learning_rate": 9.045005488474205e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008774219051701948,
+ "num_tokens": 5611971.0,
+ "step": 600
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.811292719167905,
+ "grad_norm": NaN,
+ "learning_rate": 8.825466520307355e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007863394406740553,
+ "num_tokens": 5704255.0,
+ "step": 610
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.8410104011887074,
+ "grad_norm": NaN,
+ "learning_rate": 8.605927552140506e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000578320066415472,
+ "num_tokens": 5800440.0,
+ "step": 620
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.8707280832095097,
+ "grad_norm": NaN,
+ "learning_rate": 8.386388583973656e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004275172817870043,
+ "num_tokens": 5890898.0,
+ "step": 630
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.900445765230312,
+ "grad_norm": NaN,
+ "learning_rate": 8.166849615806807e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001425291335908696,
+ "num_tokens": 5987228.0,
+ "step": 640
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.9301634472511144,
+ "grad_norm": NaN,
+ "learning_rate": 7.947310647639957e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006059414023184218,
+ "num_tokens": 6076579.0,
+ "step": 650
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.959881129271917,
+ "grad_norm": NaN,
+ "learning_rate": 7.727771679473108e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008722793892957271,
+ "num_tokens": 6171013.0,
+ "step": 660
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.9895988112927192,
+ "grad_norm": NaN,
+ "learning_rate": 7.508232711306257e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007748565993097145,
+ "num_tokens": 6265432.0,
+ "step": 670
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.0178306092124814,
+ "grad_norm": NaN,
+ "learning_rate": 7.288693743139408e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007497724390735751,
+ "num_tokens": 6349548.0,
+ "step": 680
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.0475482912332836,
+ "grad_norm": NaN,
+ "learning_rate": 7.069154774972559e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006858149892650545,
+ "num_tokens": 6440481.0,
+ "step": 690
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.0772659732540864,
+ "grad_norm": NaN,
+ "learning_rate": 6.849615806805708e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008518316331901588,
+ "num_tokens": 6533227.0,
+ "step": 700
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.1069836552748886,
+ "grad_norm": NaN,
+ "learning_rate": 6.630076838638859e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006546900600369554,
+ "num_tokens": 6631305.0,
+ "step": 710
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.136701337295691,
+ "grad_norm": NaN,
+ "learning_rate": 6.410537870472009e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006311897363048047,
+ "num_tokens": 6719619.0,
+ "step": 720
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.166419019316493,
+ "grad_norm": NaN,
+ "learning_rate": 6.19099890230516e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008787176091573201,
+ "num_tokens": 6808489.0,
+ "step": 730
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.196136701337296,
+ "grad_norm": NaN,
+ "learning_rate": 5.9714599341383095e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0010024115661508403,
+ "num_tokens": 6904544.0,
+ "step": 740
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.225854383358098,
+ "grad_norm": NaN,
+ "learning_rate": 5.75192096597146e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006462240009568632,
+ "num_tokens": 6998909.0,
+ "step": 750
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.2555720653789004,
+ "grad_norm": NaN,
+ "learning_rate": 5.532381997804611e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0005598854688287247,
+ "num_tokens": 7093305.0,
+ "step": 760
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.2852897473997027,
+ "grad_norm": NaN,
+ "learning_rate": 5.312843029637761e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00036793685212614945,
+ "num_tokens": 7184216.0,
+ "step": 770
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.315007429420505,
+ "grad_norm": NaN,
+ "learning_rate": 5.093304061470912e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000779397370206425,
+ "num_tokens": 7273139.0,
+ "step": 780
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.3447251114413077,
+ "grad_norm": NaN,
+ "learning_rate": 4.873765093304062e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006776292997528799,
+ "num_tokens": 7375877.0,
+ "step": 790
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.37444279346211,
+ "grad_norm": NaN,
+ "learning_rate": 4.654226125137213e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0013253977747808676,
+ "num_tokens": 7471802.0,
+ "step": 800
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.4041604754829122,
+ "grad_norm": NaN,
+ "learning_rate": 4.434687156970362e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007797809317708016,
+ "num_tokens": 7568137.0,
+ "step": 810
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.433878157503715,
+ "grad_norm": NaN,
+ "learning_rate": 4.215148188803513e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0005893595473025926,
+ "num_tokens": 7662880.0,
+ "step": 820
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.463595839524517,
+ "grad_norm": NaN,
+ "learning_rate": 3.995609220636663e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007197739760158584,
+ "num_tokens": 7755580.0,
+ "step": 830
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.4933135215453195,
+ "grad_norm": NaN,
+ "learning_rate": 3.776070252469814e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000664884723664727,
+ "num_tokens": 7854588.0,
+ "step": 840
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.5230312035661218,
+ "grad_norm": NaN,
+ "learning_rate": 3.5565312843029644e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000588176279416075,
+ "num_tokens": 7950417.0,
+ "step": 850
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.552748885586924,
+ "grad_norm": NaN,
+ "learning_rate": 3.3369923161361144e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007997245862497948,
+ "num_tokens": 8043286.0,
+ "step": 860
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.5824665676077267,
+ "grad_norm": NaN,
+ "learning_rate": 3.117453347969265e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0005276548559777439,
+ "num_tokens": 8138513.0,
+ "step": 870
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.612184249628529,
+ "grad_norm": NaN,
+ "learning_rate": 2.897914379802415e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00023305922804865985,
+ "num_tokens": 8228539.0,
+ "step": 880
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.6419019316493313,
+ "grad_norm": NaN,
+ "learning_rate": 2.6783754116355656e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0005359479633625597,
+ "num_tokens": 8320083.0,
+ "step": 890
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.6716196136701336,
+ "grad_norm": NaN,
+ "learning_rate": 2.458836443468716e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006636981313931755,
+ "num_tokens": 8416446.0,
+ "step": 900
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.7013372956909363,
+ "grad_norm": NaN,
+ "learning_rate": 2.239297475301866e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.002058879991818685,
+ "num_tokens": 8513551.0,
+ "step": 910
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.7310549777117386,
+ "grad_norm": NaN,
+ "learning_rate": 2.0197585071350167e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000950983911025105,
+ "num_tokens": 8603909.0,
+ "step": 920
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.760772659732541,
+ "grad_norm": NaN,
+ "learning_rate": 1.800219538968167e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0014912645376170986,
+ "num_tokens": 8695081.0,
+ "step": 930
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.790490341753343,
+ "grad_norm": NaN,
+ "learning_rate": 1.5806805708013173e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001528770748700481,
+ "num_tokens": 8790760.0,
+ "step": 940
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.8202080237741454,
+ "grad_norm": NaN,
+ "learning_rate": 1.3611416026344676e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012167295863036998,
+ "num_tokens": 8879103.0,
+ "step": 950
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.849925705794948,
+ "grad_norm": NaN,
+ "learning_rate": 1.1416026344676181e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007455661499989219,
+ "num_tokens": 8972354.0,
+ "step": 960
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.8796433878157504,
+ "grad_norm": NaN,
+ "learning_rate": 9.220636663007684e-07,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003935990913305432,
+ "num_tokens": 9069381.0,
+ "step": 970
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.9093610698365526,
+ "grad_norm": NaN,
+ "learning_rate": 7.025246981339188e-07,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00037132185607333665,
+ "num_tokens": 9164901.0,
+ "step": 980
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.9390787518573553,
+ "grad_norm": NaN,
+ "learning_rate": 4.829857299670692e-07,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00036687121028080585,
+ "num_tokens": 9258811.0,
+ "step": 990
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.9687964338781576,
+ "grad_norm": NaN,
+ "learning_rate": 2.6344676180021955e-07,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008805370052868966,
+ "num_tokens": 9351174.0,
+ "step": 1000
+ },
+ {
+ "entropy": NaN,
+ "epoch": 2.99851411589896,
+ "grad_norm": NaN,
+ "learning_rate": 4.390779363336993e-08,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007362255608313717,
+ "num_tokens": 9440239.0,
+ "step": 1010
+ }
+ ],
+ "logging_steps": 10,
+ "max_steps": 1011,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.0120089572705894e+17,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-1011/training_args.bin b/checkpoint-1011/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..737a87caa4fc2f704a1fe56a508ca4d335ca783d
--- /dev/null
+++ b/checkpoint-1011/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae0f7084660199ed7fd16027c0c6af5e726790f32a022b019f1277bec4e01142
+size 5713
diff --git a/checkpoint-125/chat_template.jinja b/checkpoint-125/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..01be9b307daa2d425f7c168c9fb145a286e0afb4
--- /dev/null
+++ b/checkpoint-125/chat_template.jinja
@@ -0,0 +1,89 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0].role == 'system' %}
+ {{- messages[0].content + '\n\n' }}
+ {%- endif %}
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0].role == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+ {%- set index = (messages|length - 1) - loop.index0 %}
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}
+ {%- set ns.multi_step_tool = false %}
+ {%- set ns.last_query_index = index %}
+ {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+ {%- if message.content is string %}
+ {%- set content = message.content %}
+ {%- else %}
+ {%- set content = '' %}
+ {%- endif %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {%- set reasoning_content = '' %}
+ {%- if message.reasoning_content is string %}
+ {%- set reasoning_content = message.reasoning_content %}
+ {%- else %}
+ {%- if '' in content %}
+ {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %}
+ {%- set content = content.split('')[-1].lstrip('\n') %}
+ {%- endif %}
+ {%- endif %}
+ {%- if loop.index0 > ns.last_query_index %}
+ {%- if loop.last or (not loop.last and reasoning_content) %}
+ {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- if message.tool_calls %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if (loop.first and content) or (not loop.first) %}
+ {{- '\n' }}
+ {%- endif %}
+ {%- if tool_call.function %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {%- if tool_call.arguments is string %}
+ {{- tool_call.arguments }}
+ {%- else %}
+ {{- tool_call.arguments | tojson }}
+ {%- endif %}
+ {{- '}\n' }}
+ {%- endfor %}
+ {%- endif %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+ {%- if enable_thinking is defined and enable_thinking is false %}
+ {{- '\n\n\n\n' }}
+ {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/checkpoint-125/config.json b/checkpoint-125/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d219d6a6d288625229a71646710b85cf7ae5def
--- /dev/null
+++ b/checkpoint-125/config.json
@@ -0,0 +1,63 @@
+{
+ "architectures": [
+ "Qwen3ForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "bfloat16",
+ "eos_token_id": 151645,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 6144,
+ "layer_types": [
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 40960,
+ "max_window_layers": 28,
+ "model_type": "qwen3",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 8,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.7.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/checkpoint-125/generation_config.json b/checkpoint-125/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec133d867b1427d43a5ee0ea332bd088fc12013
--- /dev/null
+++ b/checkpoint-125/generation_config.json
@@ -0,0 +1,12 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "temperature": 0.6,
+ "top_k": 20,
+ "top_p": 0.95,
+ "transformers_version": "5.7.0"
+}
diff --git a/checkpoint-125/model.safetensors b/checkpoint-125/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d2e0d2f571547f58479dee02aa685fc8990a4284
--- /dev/null
+++ b/checkpoint-125/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7074b51b321535051b77436f30720111854403259c583a273488e061c0d23072
+size 3441185608
diff --git a/checkpoint-125/optimizer.pt b/checkpoint-125/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c18832d9f63e5dc9e826ce8b7a6fa36249242ab6
--- /dev/null
+++ b/checkpoint-125/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b569f0710425a088cc5cd95c2def7bbb89539cdd1c3b41312c8d8fc136d2e186
+size 6882572207
diff --git a/checkpoint-125/rng_state.pth b/checkpoint-125/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5a8f17addf8aaee350625e9847751582bc9f86dd
--- /dev/null
+++ b/checkpoint-125/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61c19bab1174704a4a4441475683bf1270277af15d2e2c95e964789128e482c4
+size 14645
diff --git a/checkpoint-125/scheduler.pt b/checkpoint-125/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c0f9744d60b0805d01c098d6c1052451f23dd980
--- /dev/null
+++ b/checkpoint-125/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:770e85488b2d6f2b41a590e6c2bb599f3a174081e14b8c48d3638681b4a06597
+size 1465
diff --git a/checkpoint-125/tokenizer.json b/checkpoint-125/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7afbed2efcdf019f88ab0572ec29d3bf595dfe2
--- /dev/null
+++ b/checkpoint-125/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650
diff --git a/checkpoint-125/tokenizer_config.json b/checkpoint-125/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/checkpoint-125/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "local_files_only": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/checkpoint-125/trainer_state.json b/checkpoint-125/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e2e8da443bc02a76e800bb696414257eb8fe5e6c
--- /dev/null
+++ b/checkpoint-125/trainer_state.json
@@ -0,0 +1,154 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 1.0,
+ "eval_steps": 500,
+ "global_step": 125,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "entropy": 0.2860680317506194,
+ "epoch": 0.08,
+ "grad_norm": 7.46875,
+ "learning_rate": 1.5000000000000002e-05,
+ "loss": 0.7724831104278564,
+ "mean_token_accuracy": 0.8434554561972618,
+ "num_tokens": 94518.0,
+ "step": 10
+ },
+ {
+ "entropy": 0.3743372498080134,
+ "epoch": 0.16,
+ "grad_norm": 2.6875,
+ "learning_rate": 1.8761061946902657e-05,
+ "loss": 0.4372231483459473,
+ "mean_token_accuracy": 0.8732902690768242,
+ "num_tokens": 187445.0,
+ "step": 20
+ },
+ {
+ "entropy": 0.3575092004612088,
+ "epoch": 0.24,
+ "grad_norm": 2.625,
+ "learning_rate": 1.6991150442477876e-05,
+ "loss": 0.3412153959274292,
+ "mean_token_accuracy": 0.8936023108661175,
+ "num_tokens": 280610.0,
+ "step": 30
+ },
+ {
+ "entropy": 0.31294157058000566,
+ "epoch": 0.32,
+ "grad_norm": 2.421875,
+ "learning_rate": 1.5221238938053098e-05,
+ "loss": 0.3175334453582764,
+ "mean_token_accuracy": 0.8971926532685757,
+ "num_tokens": 379577.0,
+ "step": 40
+ },
+ {
+ "entropy": 0.30759106744080783,
+ "epoch": 0.4,
+ "grad_norm": 1.7421875,
+ "learning_rate": 1.345132743362832e-05,
+ "loss": 0.32603118419647215,
+ "mean_token_accuracy": 0.8967837899923324,
+ "num_tokens": 477577.0,
+ "step": 50
+ },
+ {
+ "entropy": 0.31574366632848977,
+ "epoch": 0.48,
+ "grad_norm": 1.9921875,
+ "learning_rate": 1.1681415929203541e-05,
+ "loss": 0.323792552947998,
+ "mean_token_accuracy": 0.8988608829677105,
+ "num_tokens": 573360.0,
+ "step": 60
+ },
+ {
+ "entropy": 0.3065364420413971,
+ "epoch": 0.56,
+ "grad_norm": 1.6171875,
+ "learning_rate": 9.911504424778762e-06,
+ "loss": 0.32266840934753416,
+ "mean_token_accuracy": 0.8985385209321975,
+ "num_tokens": 665761.0,
+ "step": 70
+ },
+ {
+ "entropy": 0.27206654362380506,
+ "epoch": 0.64,
+ "grad_norm": 2.1875,
+ "learning_rate": 8.141592920353984e-06,
+ "loss": 0.2854612350463867,
+ "mean_token_accuracy": 0.9085792370140553,
+ "num_tokens": 756921.0,
+ "step": 80
+ },
+ {
+ "entropy": 0.28962985165417193,
+ "epoch": 0.72,
+ "grad_norm": 1.8125,
+ "learning_rate": 6.371681415929204e-06,
+ "loss": 0.29041600227355957,
+ "mean_token_accuracy": 0.9059965394437313,
+ "num_tokens": 845171.0,
+ "step": 90
+ },
+ {
+ "entropy": 0.2753044681623578,
+ "epoch": 0.8,
+ "grad_norm": 2.09375,
+ "learning_rate": 4.6017699115044254e-06,
+ "loss": 0.2797576427459717,
+ "mean_token_accuracy": 0.9101039692759514,
+ "num_tokens": 939550.0,
+ "step": 100
+ },
+ {
+ "entropy": 0.2808386910706758,
+ "epoch": 0.88,
+ "grad_norm": 1.5859375,
+ "learning_rate": 2.831858407079646e-06,
+ "loss": 0.2893812894821167,
+ "mean_token_accuracy": 0.9063553236424923,
+ "num_tokens": 1041441.0,
+ "step": 110
+ },
+ {
+ "entropy": 0.2944059643894434,
+ "epoch": 0.96,
+ "grad_norm": 1.6875,
+ "learning_rate": 1.0619469026548673e-06,
+ "loss": 0.2996867895126343,
+ "mean_token_accuracy": 0.9011975660920143,
+ "num_tokens": 1132794.0,
+ "step": 120
+ }
+ ],
+ "logging_steps": 10,
+ "max_steps": 125,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 1,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.2589656731283456e+16,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-125/training_args.bin b/checkpoint-125/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0b1ceadcc46a4dbcb04d34405de43a2667755f2a
--- /dev/null
+++ b/checkpoint-125/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4e0921a533d54992c732b7cf0c8a29ef97e50bbd333d4e4c07ae7183a1feb0f
+size 5713
diff --git a/checkpoint-500/chat_template.jinja b/checkpoint-500/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..01be9b307daa2d425f7c168c9fb145a286e0afb4
--- /dev/null
+++ b/checkpoint-500/chat_template.jinja
@@ -0,0 +1,89 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0].role == 'system' %}
+ {{- messages[0].content + '\n\n' }}
+ {%- endif %}
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0].role == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+ {%- set index = (messages|length - 1) - loop.index0 %}
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}
+ {%- set ns.multi_step_tool = false %}
+ {%- set ns.last_query_index = index %}
+ {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+ {%- if message.content is string %}
+ {%- set content = message.content %}
+ {%- else %}
+ {%- set content = '' %}
+ {%- endif %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {%- set reasoning_content = '' %}
+ {%- if message.reasoning_content is string %}
+ {%- set reasoning_content = message.reasoning_content %}
+ {%- else %}
+ {%- if '' in content %}
+ {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %}
+ {%- set content = content.split('')[-1].lstrip('\n') %}
+ {%- endif %}
+ {%- endif %}
+ {%- if loop.index0 > ns.last_query_index %}
+ {%- if loop.last or (not loop.last and reasoning_content) %}
+ {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- if message.tool_calls %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if (loop.first and content) or (not loop.first) %}
+ {{- '\n' }}
+ {%- endif %}
+ {%- if tool_call.function %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {%- if tool_call.arguments is string %}
+ {{- tool_call.arguments }}
+ {%- else %}
+ {{- tool_call.arguments | tojson }}
+ {%- endif %}
+ {{- '}\n' }}
+ {%- endfor %}
+ {%- endif %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+ {%- if enable_thinking is defined and enable_thinking is false %}
+ {{- '\n\n\n\n' }}
+ {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/checkpoint-500/config.json b/checkpoint-500/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d219d6a6d288625229a71646710b85cf7ae5def
--- /dev/null
+++ b/checkpoint-500/config.json
@@ -0,0 +1,63 @@
+{
+ "architectures": [
+ "Qwen3ForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "bfloat16",
+ "eos_token_id": 151645,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 6144,
+ "layer_types": [
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 40960,
+ "max_window_layers": 28,
+ "model_type": "qwen3",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 8,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.7.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/checkpoint-500/generation_config.json b/checkpoint-500/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec133d867b1427d43a5ee0ea332bd088fc12013
--- /dev/null
+++ b/checkpoint-500/generation_config.json
@@ -0,0 +1,12 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "temperature": 0.6,
+ "top_k": 20,
+ "top_p": 0.95,
+ "transformers_version": "5.7.0"
+}
diff --git a/checkpoint-500/model.safetensors b/checkpoint-500/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..aad24a30496616243a28537371ee273f78ce5980
--- /dev/null
+++ b/checkpoint-500/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:079565c41b7390c29162b2177ece2956e65a2c94398b94f8fbdcc9aaea1e5987
+size 3441185608
diff --git a/checkpoint-500/optimizer.pt b/checkpoint-500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..90c55c866e33ae7b8a57d59da6cb54957b236309
--- /dev/null
+++ b/checkpoint-500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e11662089759b5ab33db1036a43acc423329b165030883aa32639c9ab33f035f
+size 6882572207
diff --git a/checkpoint-500/rng_state.pth b/checkpoint-500/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3105ae629835438feca5d75cb928a0254192bb7f
--- /dev/null
+++ b/checkpoint-500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f196323d7423b60f8e4ceb7dbf8715ee326c0d068e5ff164f13c63b279b9f1a0
+size 14645
diff --git a/checkpoint-500/scheduler.pt b/checkpoint-500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3cbc1d4f12d934fbe27c91c192400d02b27d2e00
--- /dev/null
+++ b/checkpoint-500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96ae61b7327de513c572cbdd449d40136fa2f4fac8357d34b3d1a844dc5168d3
+size 1465
diff --git a/checkpoint-500/tokenizer.json b/checkpoint-500/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7afbed2efcdf019f88ab0572ec29d3bf595dfe2
--- /dev/null
+++ b/checkpoint-500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650
diff --git a/checkpoint-500/tokenizer_config.json b/checkpoint-500/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/checkpoint-500/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "local_files_only": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/checkpoint-500/trainer_state.json b/checkpoint-500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..511dccfadd29552a3ad6e861310aa24e146a179e
--- /dev/null
+++ b/checkpoint-500/trainer_state.json
@@ -0,0 +1,589 @@
+{
+ "best_global_step": 500,
+ "best_metric": 0.2841346263885498,
+ "best_model_checkpoint": "/scratch/checkpoints/math_sft_v2/checkpoint-500",
+ "epoch": 1.6507018992568125,
+ "eval_steps": 100,
+ "global_step": 500,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "entropy": 0.26208699569106103,
+ "epoch": 0.03303055326176713,
+ "grad_norm": 13.3125,
+ "learning_rate": 1.5e-06,
+ "loss": 0.8376299858093261,
+ "mean_token_accuracy": 0.8421687923371792,
+ "num_tokens": 98215.0,
+ "step": 10
+ },
+ {
+ "entropy": 0.2689803125336766,
+ "epoch": 0.06606110652353427,
+ "grad_norm": 11.5625,
+ "learning_rate": 3.1666666666666667e-06,
+ "loss": 0.853145980834961,
+ "mean_token_accuracy": 0.8399689704179764,
+ "num_tokens": 191821.0,
+ "step": 20
+ },
+ {
+ "entropy": 0.28777916859835384,
+ "epoch": 0.0990916597853014,
+ "grad_norm": 12.375,
+ "learning_rate": 4.833333333333333e-06,
+ "loss": 0.7679028034210205,
+ "mean_token_accuracy": 0.8431851923465729,
+ "num_tokens": 285491.0,
+ "step": 30
+ },
+ {
+ "entropy": 0.3033197607845068,
+ "epoch": 0.13212221304706853,
+ "grad_norm": 5.375,
+ "learning_rate": 6.5000000000000004e-06,
+ "loss": 0.6137509822845459,
+ "mean_token_accuracy": 0.8549904882907867,
+ "num_tokens": 381273.0,
+ "step": 40
+ },
+ {
+ "entropy": 0.34702248610556125,
+ "epoch": 0.16515276630883569,
+ "grad_norm": 4.25,
+ "learning_rate": 8.166666666666668e-06,
+ "loss": 0.5155063152313233,
+ "mean_token_accuracy": 0.8607264116406441,
+ "num_tokens": 476481.0,
+ "step": 50
+ },
+ {
+ "entropy": 0.35821260381489994,
+ "epoch": 0.1981833195706028,
+ "grad_norm": 2.890625,
+ "learning_rate": 9.833333333333333e-06,
+ "loss": 0.42441272735595703,
+ "mean_token_accuracy": 0.8774531938135623,
+ "num_tokens": 570979.0,
+ "step": 60
+ },
+ {
+ "entropy": 0.33337059412151576,
+ "epoch": 0.23121387283236994,
+ "grad_norm": 3.15625,
+ "learning_rate": 9.835164835164835e-06,
+ "loss": 0.34101109504699706,
+ "mean_token_accuracy": 0.8939957685768605,
+ "num_tokens": 664789.0,
+ "step": 70
+ },
+ {
+ "entropy": 0.30876336991786957,
+ "epoch": 0.26424442609413706,
+ "grad_norm": 1.8515625,
+ "learning_rate": 9.652014652014653e-06,
+ "loss": 0.32317628860473635,
+ "mean_token_accuracy": 0.8987478196620942,
+ "num_tokens": 764246.0,
+ "step": 80
+ },
+ {
+ "entropy": 0.30824985895305873,
+ "epoch": 0.2972749793559042,
+ "grad_norm": 2.15625,
+ "learning_rate": 9.46886446886447e-06,
+ "loss": 0.31807353496551516,
+ "mean_token_accuracy": 0.896314675360918,
+ "num_tokens": 855689.0,
+ "step": 90
+ },
+ {
+ "entropy": 0.31522620394825934,
+ "epoch": 0.33030553261767137,
+ "grad_norm": 1.8828125,
+ "learning_rate": 9.285714285714288e-06,
+ "loss": 0.338785982131958,
+ "mean_token_accuracy": 0.89400485008955,
+ "num_tokens": 949366.0,
+ "step": 100
+ },
+ {
+ "epoch": 0.33030553261767137,
+ "eval_entropy": 0.3022541454544774,
+ "eval_loss": 0.3146008253097534,
+ "eval_mean_token_accuracy": 0.8988911988558592,
+ "eval_num_tokens": 949366.0,
+ "eval_runtime": 17.914,
+ "eval_samples_per_second": 30.088,
+ "eval_steps_per_second": 15.072,
+ "step": 100
+ },
+ {
+ "entropy": 0.2985268572345376,
+ "epoch": 0.3633360858794385,
+ "grad_norm": 2.171875,
+ "learning_rate": 9.102564102564104e-06,
+ "loss": 0.31417596340179443,
+ "mean_token_accuracy": 0.89812126532197,
+ "num_tokens": 1040854.0,
+ "step": 110
+ },
+ {
+ "entropy": 0.3040375892072916,
+ "epoch": 0.3963666391412056,
+ "grad_norm": 1.9609375,
+ "learning_rate": 8.91941391941392e-06,
+ "loss": 0.32050628662109376,
+ "mean_token_accuracy": 0.8970061101019382,
+ "num_tokens": 1132424.0,
+ "step": 120
+ },
+ {
+ "entropy": 0.3125588171184063,
+ "epoch": 0.42939719240297275,
+ "grad_norm": 1.8203125,
+ "learning_rate": 8.736263736263737e-06,
+ "loss": 0.3176266193389893,
+ "mean_token_accuracy": 0.8971590586006641,
+ "num_tokens": 1221899.0,
+ "step": 130
+ },
+ {
+ "entropy": 0.312872576713562,
+ "epoch": 0.4624277456647399,
+ "grad_norm": 1.78125,
+ "learning_rate": 8.553113553113553e-06,
+ "loss": 0.3343791484832764,
+ "mean_token_accuracy": 0.8940604299306869,
+ "num_tokens": 1315052.0,
+ "step": 140
+ },
+ {
+ "entropy": 0.27830475065857174,
+ "epoch": 0.495458298926507,
+ "grad_norm": 2.0,
+ "learning_rate": 8.369963369963371e-06,
+ "loss": 0.28718521595001223,
+ "mean_token_accuracy": 0.9051955968141556,
+ "num_tokens": 1401947.0,
+ "step": 150
+ },
+ {
+ "entropy": 0.3005422970280051,
+ "epoch": 0.5284888521882741,
+ "grad_norm": 2.34375,
+ "learning_rate": 8.186813186813188e-06,
+ "loss": 0.31569790840148926,
+ "mean_token_accuracy": 0.8999212823808194,
+ "num_tokens": 1494107.0,
+ "step": 160
+ },
+ {
+ "entropy": 0.31200682315975425,
+ "epoch": 0.5615194054500413,
+ "grad_norm": 2.015625,
+ "learning_rate": 8.003663003663006e-06,
+ "loss": 0.32325453758239747,
+ "mean_token_accuracy": 0.8952137432992459,
+ "num_tokens": 1586474.0,
+ "step": 170
+ },
+ {
+ "entropy": 0.2929861258715391,
+ "epoch": 0.5945499587118084,
+ "grad_norm": 2.09375,
+ "learning_rate": 7.820512820512822e-06,
+ "loss": 0.29693677425384524,
+ "mean_token_accuracy": 0.9031261466443539,
+ "num_tokens": 1678162.0,
+ "step": 180
+ },
+ {
+ "entropy": 0.30670158881694076,
+ "epoch": 0.6275805119735756,
+ "grad_norm": 1.828125,
+ "learning_rate": 7.637362637362638e-06,
+ "loss": 0.31567308902740476,
+ "mean_token_accuracy": 0.8983482263982296,
+ "num_tokens": 1781401.0,
+ "step": 190
+ },
+ {
+ "entropy": 0.2866023685783148,
+ "epoch": 0.6606110652353427,
+ "grad_norm": 1.6953125,
+ "learning_rate": 7.454212454212455e-06,
+ "loss": 0.29333360195159913,
+ "mean_token_accuracy": 0.904055442661047,
+ "num_tokens": 1868038.0,
+ "step": 200
+ },
+ {
+ "epoch": 0.6606110652353427,
+ "eval_entropy": 0.2921138784951634,
+ "eval_loss": 0.29292193055152893,
+ "eval_mean_token_accuracy": 0.904505420172656,
+ "eval_num_tokens": 1868038.0,
+ "eval_runtime": 17.8732,
+ "eval_samples_per_second": 30.157,
+ "eval_steps_per_second": 15.106,
+ "step": 200
+ },
+ {
+ "entropy": 0.28792863376438615,
+ "epoch": 0.6936416184971098,
+ "grad_norm": 1.9609375,
+ "learning_rate": 7.271062271062271e-06,
+ "loss": 0.29235827922821045,
+ "mean_token_accuracy": 0.9050419509410859,
+ "num_tokens": 1959819.0,
+ "step": 210
+ },
+ {
+ "entropy": 0.28098181951791046,
+ "epoch": 0.726672171758877,
+ "grad_norm": 1.8515625,
+ "learning_rate": 7.087912087912089e-06,
+ "loss": 0.2915766954421997,
+ "mean_token_accuracy": 0.9068245775997639,
+ "num_tokens": 2059113.0,
+ "step": 220
+ },
+ {
+ "entropy": 0.26028320472687483,
+ "epoch": 0.7597027250206441,
+ "grad_norm": 2.0,
+ "learning_rate": 6.9047619047619055e-06,
+ "loss": 0.2654874324798584,
+ "mean_token_accuracy": 0.9116183832287789,
+ "num_tokens": 2145321.0,
+ "step": 230
+ },
+ {
+ "entropy": 0.2855058180168271,
+ "epoch": 0.7927332782824112,
+ "grad_norm": 1.84375,
+ "learning_rate": 6.721611721611723e-06,
+ "loss": 0.2998212337493896,
+ "mean_token_accuracy": 0.9046589516103267,
+ "num_tokens": 2240859.0,
+ "step": 240
+ },
+ {
+ "entropy": 0.2783821925520897,
+ "epoch": 0.8257638315441783,
+ "grad_norm": 1.9609375,
+ "learning_rate": 6.538461538461539e-06,
+ "loss": 0.28870720863342286,
+ "mean_token_accuracy": 0.9073394626379013,
+ "num_tokens": 2333798.0,
+ "step": 250
+ },
+ {
+ "entropy": 0.277759512513876,
+ "epoch": 0.8587943848059455,
+ "grad_norm": 1.7734375,
+ "learning_rate": 6.3553113553113555e-06,
+ "loss": 0.29060728549957277,
+ "mean_token_accuracy": 0.9055352889001369,
+ "num_tokens": 2429910.0,
+ "step": 260
+ },
+ {
+ "entropy": 0.2882726760581136,
+ "epoch": 0.8918249380677127,
+ "grad_norm": 1.703125,
+ "learning_rate": 6.172161172161173e-06,
+ "loss": 0.2988502740859985,
+ "mean_token_accuracy": 0.9038613699376583,
+ "num_tokens": 2524730.0,
+ "step": 270
+ },
+ {
+ "entropy": 0.2855747708119452,
+ "epoch": 0.9248554913294798,
+ "grad_norm": 1.90625,
+ "learning_rate": 5.989010989010989e-06,
+ "loss": 0.289182186126709,
+ "mean_token_accuracy": 0.9049803033471108,
+ "num_tokens": 2615802.0,
+ "step": 280
+ },
+ {
+ "entropy": 0.26923007648438213,
+ "epoch": 0.9578860445912469,
+ "grad_norm": 2.84375,
+ "learning_rate": 5.805860805860807e-06,
+ "loss": 0.27834036350250246,
+ "mean_token_accuracy": 0.9109352678060532,
+ "num_tokens": 2709073.0,
+ "step": 290
+ },
+ {
+ "entropy": 0.2844470454379916,
+ "epoch": 0.990916597853014,
+ "grad_norm": 1.640625,
+ "learning_rate": 5.6227106227106235e-06,
+ "loss": 0.28873960971832274,
+ "mean_token_accuracy": 0.9069667845964432,
+ "num_tokens": 2805135.0,
+ "step": 300
+ },
+ {
+ "epoch": 0.990916597853014,
+ "eval_entropy": 0.285272175570329,
+ "eval_loss": 0.28705036640167236,
+ "eval_mean_token_accuracy": 0.9062359571456909,
+ "eval_num_tokens": 2805135.0,
+ "eval_runtime": 17.9038,
+ "eval_samples_per_second": 30.105,
+ "eval_steps_per_second": 15.081,
+ "step": 300
+ },
+ {
+ "entropy": 0.2869852078266633,
+ "epoch": 1.023121387283237,
+ "grad_norm": 1.9375,
+ "learning_rate": 5.43956043956044e-06,
+ "loss": 0.2850792646408081,
+ "mean_token_accuracy": 0.9064911672702203,
+ "num_tokens": 2898984.0,
+ "step": 310
+ },
+ {
+ "entropy": 0.2757460119202733,
+ "epoch": 1.0561519405450042,
+ "grad_norm": 1.8359375,
+ "learning_rate": 5.256410256410257e-06,
+ "loss": 0.2779810905456543,
+ "mean_token_accuracy": 0.9096895463764667,
+ "num_tokens": 2991694.0,
+ "step": 320
+ },
+ {
+ "entropy": 0.2600112119689584,
+ "epoch": 1.0891824938067713,
+ "grad_norm": 2.0625,
+ "learning_rate": 5.0732600732600735e-06,
+ "loss": 0.2662019729614258,
+ "mean_token_accuracy": 0.912333931028843,
+ "num_tokens": 3078862.0,
+ "step": 330
+ },
+ {
+ "entropy": 0.27342566605657337,
+ "epoch": 1.1222130470685383,
+ "grad_norm": 1.921875,
+ "learning_rate": 4.890109890109891e-06,
+ "loss": 0.28096842765808105,
+ "mean_token_accuracy": 0.9091624312102795,
+ "num_tokens": 3174256.0,
+ "step": 340
+ },
+ {
+ "entropy": 0.2671937083825469,
+ "epoch": 1.1552436003303055,
+ "grad_norm": 1.75,
+ "learning_rate": 4.706959706959707e-06,
+ "loss": 0.2723649501800537,
+ "mean_token_accuracy": 0.9101433955132961,
+ "num_tokens": 3272477.0,
+ "step": 350
+ },
+ {
+ "entropy": 0.2617026660591364,
+ "epoch": 1.1882741535920727,
+ "grad_norm": 1.9453125,
+ "learning_rate": 4.523809523809524e-06,
+ "loss": 0.26422107219696045,
+ "mean_token_accuracy": 0.913913157582283,
+ "num_tokens": 3364137.0,
+ "step": 360
+ },
+ {
+ "entropy": 0.2595691759139299,
+ "epoch": 1.2213047068538398,
+ "grad_norm": 1.5390625,
+ "learning_rate": 4.340659340659341e-06,
+ "loss": 0.2714951276779175,
+ "mean_token_accuracy": 0.911636833101511,
+ "num_tokens": 3460217.0,
+ "step": 370
+ },
+ {
+ "entropy": 0.2933320404961705,
+ "epoch": 1.254335260115607,
+ "grad_norm": 1.9765625,
+ "learning_rate": 4.157509157509158e-06,
+ "loss": 0.29573276042938235,
+ "mean_token_accuracy": 0.902192335575819,
+ "num_tokens": 3557030.0,
+ "step": 380
+ },
+ {
+ "entropy": 0.2693369995802641,
+ "epoch": 1.287365813377374,
+ "grad_norm": 1.890625,
+ "learning_rate": 3.974358974358974e-06,
+ "loss": 0.27785425186157225,
+ "mean_token_accuracy": 0.9090337291359901,
+ "num_tokens": 3654159.0,
+ "step": 390
+ },
+ {
+ "entropy": 0.2908114368095994,
+ "epoch": 1.3203963666391412,
+ "grad_norm": 2.015625,
+ "learning_rate": 3.7912087912087915e-06,
+ "loss": 0.30535905361175536,
+ "mean_token_accuracy": 0.901054035872221,
+ "num_tokens": 3749664.0,
+ "step": 400
+ },
+ {
+ "epoch": 1.3203963666391412,
+ "eval_entropy": 0.2783401116176888,
+ "eval_loss": 0.28495287895202637,
+ "eval_mean_token_accuracy": 0.9069356251645971,
+ "eval_num_tokens": 3749664.0,
+ "eval_runtime": 17.8814,
+ "eval_samples_per_second": 30.143,
+ "eval_steps_per_second": 15.099,
+ "step": 400
+ },
+ {
+ "entropy": 0.2749379588291049,
+ "epoch": 1.3534269199009084,
+ "grad_norm": 1.84375,
+ "learning_rate": 3.6080586080586083e-06,
+ "loss": 0.27795455455780027,
+ "mean_token_accuracy": 0.9090839587152004,
+ "num_tokens": 3845841.0,
+ "step": 410
+ },
+ {
+ "entropy": 0.2859474029392004,
+ "epoch": 1.3864574731626755,
+ "grad_norm": 1.6796875,
+ "learning_rate": 3.424908424908425e-06,
+ "loss": 0.2886682987213135,
+ "mean_token_accuracy": 0.9059950634837151,
+ "num_tokens": 3942402.0,
+ "step": 420
+ },
+ {
+ "entropy": 0.2537162031978369,
+ "epoch": 1.4194880264244425,
+ "grad_norm": 1.9375,
+ "learning_rate": 3.2417582417582424e-06,
+ "loss": 0.25533475875854494,
+ "mean_token_accuracy": 0.9158934533596039,
+ "num_tokens": 4034122.0,
+ "step": 430
+ },
+ {
+ "entropy": 0.28255065642297267,
+ "epoch": 1.4525185796862097,
+ "grad_norm": 1.7734375,
+ "learning_rate": 3.058608058608059e-06,
+ "loss": 0.29817631244659426,
+ "mean_token_accuracy": 0.9040707617998123,
+ "num_tokens": 4127068.0,
+ "step": 440
+ },
+ {
+ "entropy": 0.2898306529968977,
+ "epoch": 1.4855491329479769,
+ "grad_norm": 1.703125,
+ "learning_rate": 2.8754578754578755e-06,
+ "loss": 0.2923931121826172,
+ "mean_token_accuracy": 0.9030195340514183,
+ "num_tokens": 4224435.0,
+ "step": 450
+ },
+ {
+ "entropy": 0.27209411058574917,
+ "epoch": 1.518579686209744,
+ "grad_norm": 1.7578125,
+ "learning_rate": 2.6923076923076923e-06,
+ "loss": 0.28646528720855713,
+ "mean_token_accuracy": 0.9095881894230843,
+ "num_tokens": 4321028.0,
+ "step": 460
+ },
+ {
+ "entropy": 0.280727200768888,
+ "epoch": 1.5516102394715112,
+ "grad_norm": 1.7421875,
+ "learning_rate": 2.509157509157509e-06,
+ "loss": 0.2891413688659668,
+ "mean_token_accuracy": 0.9067636162042618,
+ "num_tokens": 4414916.0,
+ "step": 470
+ },
+ {
+ "entropy": 0.27109829243272543,
+ "epoch": 1.5846407927332784,
+ "grad_norm": 1.8359375,
+ "learning_rate": 2.3260073260073264e-06,
+ "loss": 0.26625516414642336,
+ "mean_token_accuracy": 0.9099020563066006,
+ "num_tokens": 4505708.0,
+ "step": 480
+ },
+ {
+ "entropy": 0.2729395367205143,
+ "epoch": 1.6176713459950454,
+ "grad_norm": 1.7890625,
+ "learning_rate": 2.1428571428571427e-06,
+ "loss": 0.2779061317443848,
+ "mean_token_accuracy": 0.909747340530157,
+ "num_tokens": 4600244.0,
+ "step": 490
+ },
+ {
+ "entropy": 0.2682301852852106,
+ "epoch": 1.6507018992568125,
+ "grad_norm": 1.6953125,
+ "learning_rate": 1.95970695970696e-06,
+ "loss": 0.274446964263916,
+ "mean_token_accuracy": 0.9094639919698239,
+ "num_tokens": 4688147.0,
+ "step": 500
+ },
+ {
+ "epoch": 1.6507018992568125,
+ "eval_entropy": 0.2756172451708052,
+ "eval_loss": 0.2841346263885498,
+ "eval_mean_token_accuracy": 0.9072470570052111,
+ "eval_num_tokens": 4688147.0,
+ "eval_runtime": 17.9119,
+ "eval_samples_per_second": 30.092,
+ "eval_steps_per_second": 15.074,
+ "step": 500
+ }
+ ],
+ "logging_steps": 10,
+ "max_steps": 606,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 2,
+ "save_steps": 100,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 5.028742352950886e+16,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-500/training_args.bin b/checkpoint-500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..133cb6661b0124bb9db0d7d208d748fd736ea689
--- /dev/null
+++ b/checkpoint-500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6696e1fef8bde5e22de2f5ae18c284f2b2d35a345157db5151571518385fb38b
+size 5713
diff --git a/checkpoint-600/chat_template.jinja b/checkpoint-600/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..01be9b307daa2d425f7c168c9fb145a286e0afb4
--- /dev/null
+++ b/checkpoint-600/chat_template.jinja
@@ -0,0 +1,89 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0].role == 'system' %}
+ {{- messages[0].content + '\n\n' }}
+ {%- endif %}
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0].role == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+ {%- set index = (messages|length - 1) - loop.index0 %}
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}
+ {%- set ns.multi_step_tool = false %}
+ {%- set ns.last_query_index = index %}
+ {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+ {%- if message.content is string %}
+ {%- set content = message.content %}
+ {%- else %}
+ {%- set content = '' %}
+ {%- endif %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {%- set reasoning_content = '' %}
+ {%- if message.reasoning_content is string %}
+ {%- set reasoning_content = message.reasoning_content %}
+ {%- else %}
+ {%- if '' in content %}
+ {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %}
+ {%- set content = content.split('')[-1].lstrip('\n') %}
+ {%- endif %}
+ {%- endif %}
+ {%- if loop.index0 > ns.last_query_index %}
+ {%- if loop.last or (not loop.last and reasoning_content) %}
+ {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- if message.tool_calls %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if (loop.first and content) or (not loop.first) %}
+ {{- '\n' }}
+ {%- endif %}
+ {%- if tool_call.function %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {%- if tool_call.arguments is string %}
+ {{- tool_call.arguments }}
+ {%- else %}
+ {{- tool_call.arguments | tojson }}
+ {%- endif %}
+ {{- '}\n' }}
+ {%- endfor %}
+ {%- endif %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+ {%- if enable_thinking is defined and enable_thinking is false %}
+ {{- '\n\n\n\n' }}
+ {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/checkpoint-600/config.json b/checkpoint-600/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d219d6a6d288625229a71646710b85cf7ae5def
--- /dev/null
+++ b/checkpoint-600/config.json
@@ -0,0 +1,63 @@
+{
+ "architectures": [
+ "Qwen3ForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "bfloat16",
+ "eos_token_id": 151645,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 6144,
+ "layer_types": [
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 40960,
+ "max_window_layers": 28,
+ "model_type": "qwen3",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 8,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.7.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/checkpoint-600/generation_config.json b/checkpoint-600/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec133d867b1427d43a5ee0ea332bd088fc12013
--- /dev/null
+++ b/checkpoint-600/generation_config.json
@@ -0,0 +1,12 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "temperature": 0.6,
+ "top_k": 20,
+ "top_p": 0.95,
+ "transformers_version": "5.7.0"
+}
diff --git a/checkpoint-600/model.safetensors b/checkpoint-600/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..888c633f416471cd0dba703e4ed15cd967deaf00
--- /dev/null
+++ b/checkpoint-600/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:066f20885c5c9a598d45fbcd21aa7ab9081dffda3fb1930d573d7c3d99b5d998
+size 3441185608
diff --git a/checkpoint-600/optimizer.pt b/checkpoint-600/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..315889bc3c6c060ec3677e282757036eb6d14155
--- /dev/null
+++ b/checkpoint-600/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c75c57fe3c14c29f7a73acb2b44e60573aaf69cd744a4ce99770cdfdbe4de45e
+size 6882572207
diff --git a/checkpoint-600/rng_state.pth b/checkpoint-600/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..1fb544a66454d577cf2a3cb06c26493972f57952
--- /dev/null
+++ b/checkpoint-600/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdc37bbd2e979f041dfbbb004a5c74bab6cdda159cb18116df728588515a9ef6
+size 14645
diff --git a/checkpoint-600/scheduler.pt b/checkpoint-600/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e2645e63119eba5c7c9361da38b055b60e6d8985
--- /dev/null
+++ b/checkpoint-600/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:540d8ab45fe72a01988766f9263355239d34a4ff0ef13ddac7f1f0d1198ce9a9
+size 1465
diff --git a/checkpoint-600/tokenizer.json b/checkpoint-600/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7afbed2efcdf019f88ab0572ec29d3bf595dfe2
--- /dev/null
+++ b/checkpoint-600/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650
diff --git a/checkpoint-600/tokenizer_config.json b/checkpoint-600/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/checkpoint-600/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "local_files_only": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/checkpoint-600/trainer_state.json b/checkpoint-600/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c4cf7cbe351f817a25fffdf26360f8e89d0ea1ce
--- /dev/null
+++ b/checkpoint-600/trainer_state.json
@@ -0,0 +1,700 @@
+{
+ "best_global_step": 600,
+ "best_metric": 0.28394654393196106,
+ "best_model_checkpoint": "/scratch/checkpoints/math_sft_v2/checkpoint-600",
+ "epoch": 1.981007431874484,
+ "eval_steps": 100,
+ "global_step": 600,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "entropy": 0.26208699569106103,
+ "epoch": 0.03303055326176713,
+ "grad_norm": 13.3125,
+ "learning_rate": 1.5e-06,
+ "loss": 0.8376299858093261,
+ "mean_token_accuracy": 0.8421687923371792,
+ "num_tokens": 98215.0,
+ "step": 10
+ },
+ {
+ "entropy": 0.2689803125336766,
+ "epoch": 0.06606110652353427,
+ "grad_norm": 11.5625,
+ "learning_rate": 3.1666666666666667e-06,
+ "loss": 0.853145980834961,
+ "mean_token_accuracy": 0.8399689704179764,
+ "num_tokens": 191821.0,
+ "step": 20
+ },
+ {
+ "entropy": 0.28777916859835384,
+ "epoch": 0.0990916597853014,
+ "grad_norm": 12.375,
+ "learning_rate": 4.833333333333333e-06,
+ "loss": 0.7679028034210205,
+ "mean_token_accuracy": 0.8431851923465729,
+ "num_tokens": 285491.0,
+ "step": 30
+ },
+ {
+ "entropy": 0.3033197607845068,
+ "epoch": 0.13212221304706853,
+ "grad_norm": 5.375,
+ "learning_rate": 6.5000000000000004e-06,
+ "loss": 0.6137509822845459,
+ "mean_token_accuracy": 0.8549904882907867,
+ "num_tokens": 381273.0,
+ "step": 40
+ },
+ {
+ "entropy": 0.34702248610556125,
+ "epoch": 0.16515276630883569,
+ "grad_norm": 4.25,
+ "learning_rate": 8.166666666666668e-06,
+ "loss": 0.5155063152313233,
+ "mean_token_accuracy": 0.8607264116406441,
+ "num_tokens": 476481.0,
+ "step": 50
+ },
+ {
+ "entropy": 0.35821260381489994,
+ "epoch": 0.1981833195706028,
+ "grad_norm": 2.890625,
+ "learning_rate": 9.833333333333333e-06,
+ "loss": 0.42441272735595703,
+ "mean_token_accuracy": 0.8774531938135623,
+ "num_tokens": 570979.0,
+ "step": 60
+ },
+ {
+ "entropy": 0.33337059412151576,
+ "epoch": 0.23121387283236994,
+ "grad_norm": 3.15625,
+ "learning_rate": 9.835164835164835e-06,
+ "loss": 0.34101109504699706,
+ "mean_token_accuracy": 0.8939957685768605,
+ "num_tokens": 664789.0,
+ "step": 70
+ },
+ {
+ "entropy": 0.30876336991786957,
+ "epoch": 0.26424442609413706,
+ "grad_norm": 1.8515625,
+ "learning_rate": 9.652014652014653e-06,
+ "loss": 0.32317628860473635,
+ "mean_token_accuracy": 0.8987478196620942,
+ "num_tokens": 764246.0,
+ "step": 80
+ },
+ {
+ "entropy": 0.30824985895305873,
+ "epoch": 0.2972749793559042,
+ "grad_norm": 2.15625,
+ "learning_rate": 9.46886446886447e-06,
+ "loss": 0.31807353496551516,
+ "mean_token_accuracy": 0.896314675360918,
+ "num_tokens": 855689.0,
+ "step": 90
+ },
+ {
+ "entropy": 0.31522620394825934,
+ "epoch": 0.33030553261767137,
+ "grad_norm": 1.8828125,
+ "learning_rate": 9.285714285714288e-06,
+ "loss": 0.338785982131958,
+ "mean_token_accuracy": 0.89400485008955,
+ "num_tokens": 949366.0,
+ "step": 100
+ },
+ {
+ "epoch": 0.33030553261767137,
+ "eval_entropy": 0.3022541454544774,
+ "eval_loss": 0.3146008253097534,
+ "eval_mean_token_accuracy": 0.8988911988558592,
+ "eval_num_tokens": 949366.0,
+ "eval_runtime": 17.914,
+ "eval_samples_per_second": 30.088,
+ "eval_steps_per_second": 15.072,
+ "step": 100
+ },
+ {
+ "entropy": 0.2985268572345376,
+ "epoch": 0.3633360858794385,
+ "grad_norm": 2.171875,
+ "learning_rate": 9.102564102564104e-06,
+ "loss": 0.31417596340179443,
+ "mean_token_accuracy": 0.89812126532197,
+ "num_tokens": 1040854.0,
+ "step": 110
+ },
+ {
+ "entropy": 0.3040375892072916,
+ "epoch": 0.3963666391412056,
+ "grad_norm": 1.9609375,
+ "learning_rate": 8.91941391941392e-06,
+ "loss": 0.32050628662109376,
+ "mean_token_accuracy": 0.8970061101019382,
+ "num_tokens": 1132424.0,
+ "step": 120
+ },
+ {
+ "entropy": 0.3125588171184063,
+ "epoch": 0.42939719240297275,
+ "grad_norm": 1.8203125,
+ "learning_rate": 8.736263736263737e-06,
+ "loss": 0.3176266193389893,
+ "mean_token_accuracy": 0.8971590586006641,
+ "num_tokens": 1221899.0,
+ "step": 130
+ },
+ {
+ "entropy": 0.312872576713562,
+ "epoch": 0.4624277456647399,
+ "grad_norm": 1.78125,
+ "learning_rate": 8.553113553113553e-06,
+ "loss": 0.3343791484832764,
+ "mean_token_accuracy": 0.8940604299306869,
+ "num_tokens": 1315052.0,
+ "step": 140
+ },
+ {
+ "entropy": 0.27830475065857174,
+ "epoch": 0.495458298926507,
+ "grad_norm": 2.0,
+ "learning_rate": 8.369963369963371e-06,
+ "loss": 0.28718521595001223,
+ "mean_token_accuracy": 0.9051955968141556,
+ "num_tokens": 1401947.0,
+ "step": 150
+ },
+ {
+ "entropy": 0.3005422970280051,
+ "epoch": 0.5284888521882741,
+ "grad_norm": 2.34375,
+ "learning_rate": 8.186813186813188e-06,
+ "loss": 0.31569790840148926,
+ "mean_token_accuracy": 0.8999212823808194,
+ "num_tokens": 1494107.0,
+ "step": 160
+ },
+ {
+ "entropy": 0.31200682315975425,
+ "epoch": 0.5615194054500413,
+ "grad_norm": 2.015625,
+ "learning_rate": 8.003663003663006e-06,
+ "loss": 0.32325453758239747,
+ "mean_token_accuracy": 0.8952137432992459,
+ "num_tokens": 1586474.0,
+ "step": 170
+ },
+ {
+ "entropy": 0.2929861258715391,
+ "epoch": 0.5945499587118084,
+ "grad_norm": 2.09375,
+ "learning_rate": 7.820512820512822e-06,
+ "loss": 0.29693677425384524,
+ "mean_token_accuracy": 0.9031261466443539,
+ "num_tokens": 1678162.0,
+ "step": 180
+ },
+ {
+ "entropy": 0.30670158881694076,
+ "epoch": 0.6275805119735756,
+ "grad_norm": 1.828125,
+ "learning_rate": 7.637362637362638e-06,
+ "loss": 0.31567308902740476,
+ "mean_token_accuracy": 0.8983482263982296,
+ "num_tokens": 1781401.0,
+ "step": 190
+ },
+ {
+ "entropy": 0.2866023685783148,
+ "epoch": 0.6606110652353427,
+ "grad_norm": 1.6953125,
+ "learning_rate": 7.454212454212455e-06,
+ "loss": 0.29333360195159913,
+ "mean_token_accuracy": 0.904055442661047,
+ "num_tokens": 1868038.0,
+ "step": 200
+ },
+ {
+ "epoch": 0.6606110652353427,
+ "eval_entropy": 0.2921138784951634,
+ "eval_loss": 0.29292193055152893,
+ "eval_mean_token_accuracy": 0.904505420172656,
+ "eval_num_tokens": 1868038.0,
+ "eval_runtime": 17.8732,
+ "eval_samples_per_second": 30.157,
+ "eval_steps_per_second": 15.106,
+ "step": 200
+ },
+ {
+ "entropy": 0.28792863376438615,
+ "epoch": 0.6936416184971098,
+ "grad_norm": 1.9609375,
+ "learning_rate": 7.271062271062271e-06,
+ "loss": 0.29235827922821045,
+ "mean_token_accuracy": 0.9050419509410859,
+ "num_tokens": 1959819.0,
+ "step": 210
+ },
+ {
+ "entropy": 0.28098181951791046,
+ "epoch": 0.726672171758877,
+ "grad_norm": 1.8515625,
+ "learning_rate": 7.087912087912089e-06,
+ "loss": 0.2915766954421997,
+ "mean_token_accuracy": 0.9068245775997639,
+ "num_tokens": 2059113.0,
+ "step": 220
+ },
+ {
+ "entropy": 0.26028320472687483,
+ "epoch": 0.7597027250206441,
+ "grad_norm": 2.0,
+ "learning_rate": 6.9047619047619055e-06,
+ "loss": 0.2654874324798584,
+ "mean_token_accuracy": 0.9116183832287789,
+ "num_tokens": 2145321.0,
+ "step": 230
+ },
+ {
+ "entropy": 0.2855058180168271,
+ "epoch": 0.7927332782824112,
+ "grad_norm": 1.84375,
+ "learning_rate": 6.721611721611723e-06,
+ "loss": 0.2998212337493896,
+ "mean_token_accuracy": 0.9046589516103267,
+ "num_tokens": 2240859.0,
+ "step": 240
+ },
+ {
+ "entropy": 0.2783821925520897,
+ "epoch": 0.8257638315441783,
+ "grad_norm": 1.9609375,
+ "learning_rate": 6.538461538461539e-06,
+ "loss": 0.28870720863342286,
+ "mean_token_accuracy": 0.9073394626379013,
+ "num_tokens": 2333798.0,
+ "step": 250
+ },
+ {
+ "entropy": 0.277759512513876,
+ "epoch": 0.8587943848059455,
+ "grad_norm": 1.7734375,
+ "learning_rate": 6.3553113553113555e-06,
+ "loss": 0.29060728549957277,
+ "mean_token_accuracy": 0.9055352889001369,
+ "num_tokens": 2429910.0,
+ "step": 260
+ },
+ {
+ "entropy": 0.2882726760581136,
+ "epoch": 0.8918249380677127,
+ "grad_norm": 1.703125,
+ "learning_rate": 6.172161172161173e-06,
+ "loss": 0.2988502740859985,
+ "mean_token_accuracy": 0.9038613699376583,
+ "num_tokens": 2524730.0,
+ "step": 270
+ },
+ {
+ "entropy": 0.2855747708119452,
+ "epoch": 0.9248554913294798,
+ "grad_norm": 1.90625,
+ "learning_rate": 5.989010989010989e-06,
+ "loss": 0.289182186126709,
+ "mean_token_accuracy": 0.9049803033471108,
+ "num_tokens": 2615802.0,
+ "step": 280
+ },
+ {
+ "entropy": 0.26923007648438213,
+ "epoch": 0.9578860445912469,
+ "grad_norm": 2.84375,
+ "learning_rate": 5.805860805860807e-06,
+ "loss": 0.27834036350250246,
+ "mean_token_accuracy": 0.9109352678060532,
+ "num_tokens": 2709073.0,
+ "step": 290
+ },
+ {
+ "entropy": 0.2844470454379916,
+ "epoch": 0.990916597853014,
+ "grad_norm": 1.640625,
+ "learning_rate": 5.6227106227106235e-06,
+ "loss": 0.28873960971832274,
+ "mean_token_accuracy": 0.9069667845964432,
+ "num_tokens": 2805135.0,
+ "step": 300
+ },
+ {
+ "epoch": 0.990916597853014,
+ "eval_entropy": 0.285272175570329,
+ "eval_loss": 0.28705036640167236,
+ "eval_mean_token_accuracy": 0.9062359571456909,
+ "eval_num_tokens": 2805135.0,
+ "eval_runtime": 17.9038,
+ "eval_samples_per_second": 30.105,
+ "eval_steps_per_second": 15.081,
+ "step": 300
+ },
+ {
+ "entropy": 0.2869852078266633,
+ "epoch": 1.023121387283237,
+ "grad_norm": 1.9375,
+ "learning_rate": 5.43956043956044e-06,
+ "loss": 0.2850792646408081,
+ "mean_token_accuracy": 0.9064911672702203,
+ "num_tokens": 2898984.0,
+ "step": 310
+ },
+ {
+ "entropy": 0.2757460119202733,
+ "epoch": 1.0561519405450042,
+ "grad_norm": 1.8359375,
+ "learning_rate": 5.256410256410257e-06,
+ "loss": 0.2779810905456543,
+ "mean_token_accuracy": 0.9096895463764667,
+ "num_tokens": 2991694.0,
+ "step": 320
+ },
+ {
+ "entropy": 0.2600112119689584,
+ "epoch": 1.0891824938067713,
+ "grad_norm": 2.0625,
+ "learning_rate": 5.0732600732600735e-06,
+ "loss": 0.2662019729614258,
+ "mean_token_accuracy": 0.912333931028843,
+ "num_tokens": 3078862.0,
+ "step": 330
+ },
+ {
+ "entropy": 0.27342566605657337,
+ "epoch": 1.1222130470685383,
+ "grad_norm": 1.921875,
+ "learning_rate": 4.890109890109891e-06,
+ "loss": 0.28096842765808105,
+ "mean_token_accuracy": 0.9091624312102795,
+ "num_tokens": 3174256.0,
+ "step": 340
+ },
+ {
+ "entropy": 0.2671937083825469,
+ "epoch": 1.1552436003303055,
+ "grad_norm": 1.75,
+ "learning_rate": 4.706959706959707e-06,
+ "loss": 0.2723649501800537,
+ "mean_token_accuracy": 0.9101433955132961,
+ "num_tokens": 3272477.0,
+ "step": 350
+ },
+ {
+ "entropy": 0.2617026660591364,
+ "epoch": 1.1882741535920727,
+ "grad_norm": 1.9453125,
+ "learning_rate": 4.523809523809524e-06,
+ "loss": 0.26422107219696045,
+ "mean_token_accuracy": 0.913913157582283,
+ "num_tokens": 3364137.0,
+ "step": 360
+ },
+ {
+ "entropy": 0.2595691759139299,
+ "epoch": 1.2213047068538398,
+ "grad_norm": 1.5390625,
+ "learning_rate": 4.340659340659341e-06,
+ "loss": 0.2714951276779175,
+ "mean_token_accuracy": 0.911636833101511,
+ "num_tokens": 3460217.0,
+ "step": 370
+ },
+ {
+ "entropy": 0.2933320404961705,
+ "epoch": 1.254335260115607,
+ "grad_norm": 1.9765625,
+ "learning_rate": 4.157509157509158e-06,
+ "loss": 0.29573276042938235,
+ "mean_token_accuracy": 0.902192335575819,
+ "num_tokens": 3557030.0,
+ "step": 380
+ },
+ {
+ "entropy": 0.2693369995802641,
+ "epoch": 1.287365813377374,
+ "grad_norm": 1.890625,
+ "learning_rate": 3.974358974358974e-06,
+ "loss": 0.27785425186157225,
+ "mean_token_accuracy": 0.9090337291359901,
+ "num_tokens": 3654159.0,
+ "step": 390
+ },
+ {
+ "entropy": 0.2908114368095994,
+ "epoch": 1.3203963666391412,
+ "grad_norm": 2.015625,
+ "learning_rate": 3.7912087912087915e-06,
+ "loss": 0.30535905361175536,
+ "mean_token_accuracy": 0.901054035872221,
+ "num_tokens": 3749664.0,
+ "step": 400
+ },
+ {
+ "epoch": 1.3203963666391412,
+ "eval_entropy": 0.2783401116176888,
+ "eval_loss": 0.28495287895202637,
+ "eval_mean_token_accuracy": 0.9069356251645971,
+ "eval_num_tokens": 3749664.0,
+ "eval_runtime": 17.8814,
+ "eval_samples_per_second": 30.143,
+ "eval_steps_per_second": 15.099,
+ "step": 400
+ },
+ {
+ "entropy": 0.2749379588291049,
+ "epoch": 1.3534269199009084,
+ "grad_norm": 1.84375,
+ "learning_rate": 3.6080586080586083e-06,
+ "loss": 0.27795455455780027,
+ "mean_token_accuracy": 0.9090839587152004,
+ "num_tokens": 3845841.0,
+ "step": 410
+ },
+ {
+ "entropy": 0.2859474029392004,
+ "epoch": 1.3864574731626755,
+ "grad_norm": 1.6796875,
+ "learning_rate": 3.424908424908425e-06,
+ "loss": 0.2886682987213135,
+ "mean_token_accuracy": 0.9059950634837151,
+ "num_tokens": 3942402.0,
+ "step": 420
+ },
+ {
+ "entropy": 0.2537162031978369,
+ "epoch": 1.4194880264244425,
+ "grad_norm": 1.9375,
+ "learning_rate": 3.2417582417582424e-06,
+ "loss": 0.25533475875854494,
+ "mean_token_accuracy": 0.9158934533596039,
+ "num_tokens": 4034122.0,
+ "step": 430
+ },
+ {
+ "entropy": 0.28255065642297267,
+ "epoch": 1.4525185796862097,
+ "grad_norm": 1.7734375,
+ "learning_rate": 3.058608058608059e-06,
+ "loss": 0.29817631244659426,
+ "mean_token_accuracy": 0.9040707617998123,
+ "num_tokens": 4127068.0,
+ "step": 440
+ },
+ {
+ "entropy": 0.2898306529968977,
+ "epoch": 1.4855491329479769,
+ "grad_norm": 1.703125,
+ "learning_rate": 2.8754578754578755e-06,
+ "loss": 0.2923931121826172,
+ "mean_token_accuracy": 0.9030195340514183,
+ "num_tokens": 4224435.0,
+ "step": 450
+ },
+ {
+ "entropy": 0.27209411058574917,
+ "epoch": 1.518579686209744,
+ "grad_norm": 1.7578125,
+ "learning_rate": 2.6923076923076923e-06,
+ "loss": 0.28646528720855713,
+ "mean_token_accuracy": 0.9095881894230843,
+ "num_tokens": 4321028.0,
+ "step": 460
+ },
+ {
+ "entropy": 0.280727200768888,
+ "epoch": 1.5516102394715112,
+ "grad_norm": 1.7421875,
+ "learning_rate": 2.509157509157509e-06,
+ "loss": 0.2891413688659668,
+ "mean_token_accuracy": 0.9067636162042618,
+ "num_tokens": 4414916.0,
+ "step": 470
+ },
+ {
+ "entropy": 0.27109829243272543,
+ "epoch": 1.5846407927332784,
+ "grad_norm": 1.8359375,
+ "learning_rate": 2.3260073260073264e-06,
+ "loss": 0.26625516414642336,
+ "mean_token_accuracy": 0.9099020563066006,
+ "num_tokens": 4505708.0,
+ "step": 480
+ },
+ {
+ "entropy": 0.2729395367205143,
+ "epoch": 1.6176713459950454,
+ "grad_norm": 1.7890625,
+ "learning_rate": 2.1428571428571427e-06,
+ "loss": 0.2779061317443848,
+ "mean_token_accuracy": 0.909747340530157,
+ "num_tokens": 4600244.0,
+ "step": 490
+ },
+ {
+ "entropy": 0.2682301852852106,
+ "epoch": 1.6507018992568125,
+ "grad_norm": 1.6953125,
+ "learning_rate": 1.95970695970696e-06,
+ "loss": 0.274446964263916,
+ "mean_token_accuracy": 0.9094639919698239,
+ "num_tokens": 4688147.0,
+ "step": 500
+ },
+ {
+ "epoch": 1.6507018992568125,
+ "eval_entropy": 0.2756172451708052,
+ "eval_loss": 0.2841346263885498,
+ "eval_mean_token_accuracy": 0.9072470570052111,
+ "eval_num_tokens": 4688147.0,
+ "eval_runtime": 17.9119,
+ "eval_samples_per_second": 30.092,
+ "eval_steps_per_second": 15.074,
+ "step": 500
+ },
+ {
+ "entropy": 0.26342453975230457,
+ "epoch": 1.6837324525185797,
+ "grad_norm": 1.890625,
+ "learning_rate": 1.7765567765567768e-06,
+ "loss": 0.26432271003723146,
+ "mean_token_accuracy": 0.9114409819245338,
+ "num_tokens": 4780124.0,
+ "step": 510
+ },
+ {
+ "entropy": 0.28657868495211003,
+ "epoch": 1.7167630057803467,
+ "grad_norm": 1.7734375,
+ "learning_rate": 1.5934065934065933e-06,
+ "loss": 0.2983593225479126,
+ "mean_token_accuracy": 0.9028328903019428,
+ "num_tokens": 4874633.0,
+ "step": 520
+ },
+ {
+ "entropy": 0.25633321776986123,
+ "epoch": 1.7497935590421139,
+ "grad_norm": 1.8359375,
+ "learning_rate": 1.4102564102564104e-06,
+ "loss": 0.2606488227844238,
+ "mean_token_accuracy": 0.9153999648988247,
+ "num_tokens": 4967263.0,
+ "step": 530
+ },
+ {
+ "entropy": 0.26120030507445335,
+ "epoch": 1.782824112303881,
+ "grad_norm": 2.078125,
+ "learning_rate": 1.2271062271062271e-06,
+ "loss": 0.2706313610076904,
+ "mean_token_accuracy": 0.9130682416260243,
+ "num_tokens": 5056736.0,
+ "step": 540
+ },
+ {
+ "entropy": 0.27009565513581035,
+ "epoch": 1.8158546655656482,
+ "grad_norm": 1.8828125,
+ "learning_rate": 1.0439560439560442e-06,
+ "loss": 0.277858567237854,
+ "mean_token_accuracy": 0.9087999157607556,
+ "num_tokens": 5150617.0,
+ "step": 550
+ },
+ {
+ "entropy": 0.2819518899545074,
+ "epoch": 1.8488852188274154,
+ "grad_norm": 1.9609375,
+ "learning_rate": 8.608058608058609e-07,
+ "loss": 0.28743035793304444,
+ "mean_token_accuracy": 0.9067244723439216,
+ "num_tokens": 5243538.0,
+ "step": 560
+ },
+ {
+ "entropy": 0.2858045116066933,
+ "epoch": 1.8819157720891826,
+ "grad_norm": 2.203125,
+ "learning_rate": 6.776556776556777e-07,
+ "loss": 0.2939825296401978,
+ "mean_token_accuracy": 0.904579871147871,
+ "num_tokens": 5331441.0,
+ "step": 570
+ },
+ {
+ "entropy": 0.2926233192905784,
+ "epoch": 1.9149463253509498,
+ "grad_norm": 2.3125,
+ "learning_rate": 4.945054945054946e-07,
+ "loss": 0.29961700439453126,
+ "mean_token_accuracy": 0.9033655568957328,
+ "num_tokens": 5426734.0,
+ "step": 580
+ },
+ {
+ "entropy": 0.2706678254529834,
+ "epoch": 1.9479768786127167,
+ "grad_norm": 2.03125,
+ "learning_rate": 3.113553113553114e-07,
+ "loss": 0.27953526973724363,
+ "mean_token_accuracy": 0.9093762949109078,
+ "num_tokens": 5518190.0,
+ "step": 590
+ },
+ {
+ "entropy": 0.27056904807686805,
+ "epoch": 1.981007431874484,
+ "grad_norm": 2.09375,
+ "learning_rate": 1.282051282051282e-07,
+ "loss": 0.2764536619186401,
+ "mean_token_accuracy": 0.9110228553414345,
+ "num_tokens": 5609238.0,
+ "step": 600
+ },
+ {
+ "epoch": 1.981007431874484,
+ "eval_entropy": 0.2757842331572815,
+ "eval_loss": 0.28394654393196106,
+ "eval_mean_token_accuracy": 0.9073928307603907,
+ "eval_num_tokens": 5609238.0,
+ "eval_runtime": 18.1138,
+ "eval_samples_per_second": 29.756,
+ "eval_steps_per_second": 14.906,
+ "step": 600
+ }
+ ],
+ "logging_steps": 10,
+ "max_steps": 606,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 2,
+ "save_steps": 100,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 6.018332557481165e+16,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-600/training_args.bin b/checkpoint-600/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..133cb6661b0124bb9db0d7d208d748fd736ea689
--- /dev/null
+++ b/checkpoint-600/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6696e1fef8bde5e22de2f5ae18c284f2b2d35a345157db5151571518385fb38b
+size 5713
diff --git a/checkpoint-606/chat_template.jinja b/checkpoint-606/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..01be9b307daa2d425f7c168c9fb145a286e0afb4
--- /dev/null
+++ b/checkpoint-606/chat_template.jinja
@@ -0,0 +1,89 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0].role == 'system' %}
+ {{- messages[0].content + '\n\n' }}
+ {%- endif %}
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0].role == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+ {%- set index = (messages|length - 1) - loop.index0 %}
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}
+ {%- set ns.multi_step_tool = false %}
+ {%- set ns.last_query_index = index %}
+ {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+ {%- if message.content is string %}
+ {%- set content = message.content %}
+ {%- else %}
+ {%- set content = '' %}
+ {%- endif %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {%- set reasoning_content = '' %}
+ {%- if message.reasoning_content is string %}
+ {%- set reasoning_content = message.reasoning_content %}
+ {%- else %}
+ {%- if '' in content %}
+ {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %}
+ {%- set content = content.split('')[-1].lstrip('\n') %}
+ {%- endif %}
+ {%- endif %}
+ {%- if loop.index0 > ns.last_query_index %}
+ {%- if loop.last or (not loop.last and reasoning_content) %}
+ {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- if message.tool_calls %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if (loop.first and content) or (not loop.first) %}
+ {{- '\n' }}
+ {%- endif %}
+ {%- if tool_call.function %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {%- if tool_call.arguments is string %}
+ {{- tool_call.arguments }}
+ {%- else %}
+ {{- tool_call.arguments | tojson }}
+ {%- endif %}
+ {{- '}\n' }}
+ {%- endfor %}
+ {%- endif %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+ {%- if enable_thinking is defined and enable_thinking is false %}
+ {{- '\n\n\n\n' }}
+ {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/checkpoint-606/config.json b/checkpoint-606/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d219d6a6d288625229a71646710b85cf7ae5def
--- /dev/null
+++ b/checkpoint-606/config.json
@@ -0,0 +1,63 @@
+{
+ "architectures": [
+ "Qwen3ForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "bfloat16",
+ "eos_token_id": 151645,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 6144,
+ "layer_types": [
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 40960,
+ "max_window_layers": 28,
+ "model_type": "qwen3",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 8,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.7.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/checkpoint-606/generation_config.json b/checkpoint-606/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec133d867b1427d43a5ee0ea332bd088fc12013
--- /dev/null
+++ b/checkpoint-606/generation_config.json
@@ -0,0 +1,12 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "temperature": 0.6,
+ "top_k": 20,
+ "top_p": 0.95,
+ "transformers_version": "5.7.0"
+}
diff --git a/checkpoint-606/model.safetensors b/checkpoint-606/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..b08d1501dfdc93522108efc90dd1cd038d65cc7a
--- /dev/null
+++ b/checkpoint-606/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2e0bbbd38e2572fcb7a7cdd2bf58165f2cb2a115544c68b132848cd7973d106
+size 3441185608
diff --git a/checkpoint-606/optimizer.pt b/checkpoint-606/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..590b9fb7b131560c1b58842b49379b273c5a6449
--- /dev/null
+++ b/checkpoint-606/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc3d8578da8d88239c2024e2dd1da1012e2173f003e746efcc15b17915d0a03c
+size 6882572207
diff --git a/checkpoint-606/rng_state.pth b/checkpoint-606/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..eed58b88838bddfd46b410984c9cbddf403b1437
--- /dev/null
+++ b/checkpoint-606/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:363c5df1543d2c82b2f13164f35bdd0367ceb32e7fa1b2f67c19df073a08b17b
+size 14645
diff --git a/checkpoint-606/scheduler.pt b/checkpoint-606/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f182dc5e2546b5fdccbdc22b90eb16a565c67e6c
--- /dev/null
+++ b/checkpoint-606/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef108e31778ee002976f58f4a1ef7b3f1228ac6f0b9152b3822e08d82d99376b
+size 1465
diff --git a/checkpoint-606/tokenizer.json b/checkpoint-606/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7afbed2efcdf019f88ab0572ec29d3bf595dfe2
--- /dev/null
+++ b/checkpoint-606/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650
diff --git a/checkpoint-606/tokenizer_config.json b/checkpoint-606/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/checkpoint-606/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "local_files_only": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/checkpoint-606/trainer_state.json b/checkpoint-606/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a4ec67787e726511e6fbf586ff5bd33e585236d7
--- /dev/null
+++ b/checkpoint-606/trainer_state.json
@@ -0,0 +1,711 @@
+{
+ "best_global_step": 600,
+ "best_metric": 0.28394654393196106,
+ "best_model_checkpoint": "/scratch/checkpoints/math_sft_v2/checkpoint-600",
+ "epoch": 2.0,
+ "eval_steps": 100,
+ "global_step": 606,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "entropy": 0.26208699569106103,
+ "epoch": 0.03303055326176713,
+ "grad_norm": 13.3125,
+ "learning_rate": 1.5e-06,
+ "loss": 0.8376299858093261,
+ "mean_token_accuracy": 0.8421687923371792,
+ "num_tokens": 98215.0,
+ "step": 10
+ },
+ {
+ "entropy": 0.2689803125336766,
+ "epoch": 0.06606110652353427,
+ "grad_norm": 11.5625,
+ "learning_rate": 3.1666666666666667e-06,
+ "loss": 0.853145980834961,
+ "mean_token_accuracy": 0.8399689704179764,
+ "num_tokens": 191821.0,
+ "step": 20
+ },
+ {
+ "entropy": 0.28777916859835384,
+ "epoch": 0.0990916597853014,
+ "grad_norm": 12.375,
+ "learning_rate": 4.833333333333333e-06,
+ "loss": 0.7679028034210205,
+ "mean_token_accuracy": 0.8431851923465729,
+ "num_tokens": 285491.0,
+ "step": 30
+ },
+ {
+ "entropy": 0.3033197607845068,
+ "epoch": 0.13212221304706853,
+ "grad_norm": 5.375,
+ "learning_rate": 6.5000000000000004e-06,
+ "loss": 0.6137509822845459,
+ "mean_token_accuracy": 0.8549904882907867,
+ "num_tokens": 381273.0,
+ "step": 40
+ },
+ {
+ "entropy": 0.34702248610556125,
+ "epoch": 0.16515276630883569,
+ "grad_norm": 4.25,
+ "learning_rate": 8.166666666666668e-06,
+ "loss": 0.5155063152313233,
+ "mean_token_accuracy": 0.8607264116406441,
+ "num_tokens": 476481.0,
+ "step": 50
+ },
+ {
+ "entropy": 0.35821260381489994,
+ "epoch": 0.1981833195706028,
+ "grad_norm": 2.890625,
+ "learning_rate": 9.833333333333333e-06,
+ "loss": 0.42441272735595703,
+ "mean_token_accuracy": 0.8774531938135623,
+ "num_tokens": 570979.0,
+ "step": 60
+ },
+ {
+ "entropy": 0.33337059412151576,
+ "epoch": 0.23121387283236994,
+ "grad_norm": 3.15625,
+ "learning_rate": 9.835164835164835e-06,
+ "loss": 0.34101109504699706,
+ "mean_token_accuracy": 0.8939957685768605,
+ "num_tokens": 664789.0,
+ "step": 70
+ },
+ {
+ "entropy": 0.30876336991786957,
+ "epoch": 0.26424442609413706,
+ "grad_norm": 1.8515625,
+ "learning_rate": 9.652014652014653e-06,
+ "loss": 0.32317628860473635,
+ "mean_token_accuracy": 0.8987478196620942,
+ "num_tokens": 764246.0,
+ "step": 80
+ },
+ {
+ "entropy": 0.30824985895305873,
+ "epoch": 0.2972749793559042,
+ "grad_norm": 2.15625,
+ "learning_rate": 9.46886446886447e-06,
+ "loss": 0.31807353496551516,
+ "mean_token_accuracy": 0.896314675360918,
+ "num_tokens": 855689.0,
+ "step": 90
+ },
+ {
+ "entropy": 0.31522620394825934,
+ "epoch": 0.33030553261767137,
+ "grad_norm": 1.8828125,
+ "learning_rate": 9.285714285714288e-06,
+ "loss": 0.338785982131958,
+ "mean_token_accuracy": 0.89400485008955,
+ "num_tokens": 949366.0,
+ "step": 100
+ },
+ {
+ "epoch": 0.33030553261767137,
+ "eval_entropy": 0.3022541454544774,
+ "eval_loss": 0.3146008253097534,
+ "eval_mean_token_accuracy": 0.8988911988558592,
+ "eval_num_tokens": 949366.0,
+ "eval_runtime": 17.914,
+ "eval_samples_per_second": 30.088,
+ "eval_steps_per_second": 15.072,
+ "step": 100
+ },
+ {
+ "entropy": 0.2985268572345376,
+ "epoch": 0.3633360858794385,
+ "grad_norm": 2.171875,
+ "learning_rate": 9.102564102564104e-06,
+ "loss": 0.31417596340179443,
+ "mean_token_accuracy": 0.89812126532197,
+ "num_tokens": 1040854.0,
+ "step": 110
+ },
+ {
+ "entropy": 0.3040375892072916,
+ "epoch": 0.3963666391412056,
+ "grad_norm": 1.9609375,
+ "learning_rate": 8.91941391941392e-06,
+ "loss": 0.32050628662109376,
+ "mean_token_accuracy": 0.8970061101019382,
+ "num_tokens": 1132424.0,
+ "step": 120
+ },
+ {
+ "entropy": 0.3125588171184063,
+ "epoch": 0.42939719240297275,
+ "grad_norm": 1.8203125,
+ "learning_rate": 8.736263736263737e-06,
+ "loss": 0.3176266193389893,
+ "mean_token_accuracy": 0.8971590586006641,
+ "num_tokens": 1221899.0,
+ "step": 130
+ },
+ {
+ "entropy": 0.312872576713562,
+ "epoch": 0.4624277456647399,
+ "grad_norm": 1.78125,
+ "learning_rate": 8.553113553113553e-06,
+ "loss": 0.3343791484832764,
+ "mean_token_accuracy": 0.8940604299306869,
+ "num_tokens": 1315052.0,
+ "step": 140
+ },
+ {
+ "entropy": 0.27830475065857174,
+ "epoch": 0.495458298926507,
+ "grad_norm": 2.0,
+ "learning_rate": 8.369963369963371e-06,
+ "loss": 0.28718521595001223,
+ "mean_token_accuracy": 0.9051955968141556,
+ "num_tokens": 1401947.0,
+ "step": 150
+ },
+ {
+ "entropy": 0.3005422970280051,
+ "epoch": 0.5284888521882741,
+ "grad_norm": 2.34375,
+ "learning_rate": 8.186813186813188e-06,
+ "loss": 0.31569790840148926,
+ "mean_token_accuracy": 0.8999212823808194,
+ "num_tokens": 1494107.0,
+ "step": 160
+ },
+ {
+ "entropy": 0.31200682315975425,
+ "epoch": 0.5615194054500413,
+ "grad_norm": 2.015625,
+ "learning_rate": 8.003663003663006e-06,
+ "loss": 0.32325453758239747,
+ "mean_token_accuracy": 0.8952137432992459,
+ "num_tokens": 1586474.0,
+ "step": 170
+ },
+ {
+ "entropy": 0.2929861258715391,
+ "epoch": 0.5945499587118084,
+ "grad_norm": 2.09375,
+ "learning_rate": 7.820512820512822e-06,
+ "loss": 0.29693677425384524,
+ "mean_token_accuracy": 0.9031261466443539,
+ "num_tokens": 1678162.0,
+ "step": 180
+ },
+ {
+ "entropy": 0.30670158881694076,
+ "epoch": 0.6275805119735756,
+ "grad_norm": 1.828125,
+ "learning_rate": 7.637362637362638e-06,
+ "loss": 0.31567308902740476,
+ "mean_token_accuracy": 0.8983482263982296,
+ "num_tokens": 1781401.0,
+ "step": 190
+ },
+ {
+ "entropy": 0.2866023685783148,
+ "epoch": 0.6606110652353427,
+ "grad_norm": 1.6953125,
+ "learning_rate": 7.454212454212455e-06,
+ "loss": 0.29333360195159913,
+ "mean_token_accuracy": 0.904055442661047,
+ "num_tokens": 1868038.0,
+ "step": 200
+ },
+ {
+ "epoch": 0.6606110652353427,
+ "eval_entropy": 0.2921138784951634,
+ "eval_loss": 0.29292193055152893,
+ "eval_mean_token_accuracy": 0.904505420172656,
+ "eval_num_tokens": 1868038.0,
+ "eval_runtime": 17.8732,
+ "eval_samples_per_second": 30.157,
+ "eval_steps_per_second": 15.106,
+ "step": 200
+ },
+ {
+ "entropy": 0.28792863376438615,
+ "epoch": 0.6936416184971098,
+ "grad_norm": 1.9609375,
+ "learning_rate": 7.271062271062271e-06,
+ "loss": 0.29235827922821045,
+ "mean_token_accuracy": 0.9050419509410859,
+ "num_tokens": 1959819.0,
+ "step": 210
+ },
+ {
+ "entropy": 0.28098181951791046,
+ "epoch": 0.726672171758877,
+ "grad_norm": 1.8515625,
+ "learning_rate": 7.087912087912089e-06,
+ "loss": 0.2915766954421997,
+ "mean_token_accuracy": 0.9068245775997639,
+ "num_tokens": 2059113.0,
+ "step": 220
+ },
+ {
+ "entropy": 0.26028320472687483,
+ "epoch": 0.7597027250206441,
+ "grad_norm": 2.0,
+ "learning_rate": 6.9047619047619055e-06,
+ "loss": 0.2654874324798584,
+ "mean_token_accuracy": 0.9116183832287789,
+ "num_tokens": 2145321.0,
+ "step": 230
+ },
+ {
+ "entropy": 0.2855058180168271,
+ "epoch": 0.7927332782824112,
+ "grad_norm": 1.84375,
+ "learning_rate": 6.721611721611723e-06,
+ "loss": 0.2998212337493896,
+ "mean_token_accuracy": 0.9046589516103267,
+ "num_tokens": 2240859.0,
+ "step": 240
+ },
+ {
+ "entropy": 0.2783821925520897,
+ "epoch": 0.8257638315441783,
+ "grad_norm": 1.9609375,
+ "learning_rate": 6.538461538461539e-06,
+ "loss": 0.28870720863342286,
+ "mean_token_accuracy": 0.9073394626379013,
+ "num_tokens": 2333798.0,
+ "step": 250
+ },
+ {
+ "entropy": 0.277759512513876,
+ "epoch": 0.8587943848059455,
+ "grad_norm": 1.7734375,
+ "learning_rate": 6.3553113553113555e-06,
+ "loss": 0.29060728549957277,
+ "mean_token_accuracy": 0.9055352889001369,
+ "num_tokens": 2429910.0,
+ "step": 260
+ },
+ {
+ "entropy": 0.2882726760581136,
+ "epoch": 0.8918249380677127,
+ "grad_norm": 1.703125,
+ "learning_rate": 6.172161172161173e-06,
+ "loss": 0.2988502740859985,
+ "mean_token_accuracy": 0.9038613699376583,
+ "num_tokens": 2524730.0,
+ "step": 270
+ },
+ {
+ "entropy": 0.2855747708119452,
+ "epoch": 0.9248554913294798,
+ "grad_norm": 1.90625,
+ "learning_rate": 5.989010989010989e-06,
+ "loss": 0.289182186126709,
+ "mean_token_accuracy": 0.9049803033471108,
+ "num_tokens": 2615802.0,
+ "step": 280
+ },
+ {
+ "entropy": 0.26923007648438213,
+ "epoch": 0.9578860445912469,
+ "grad_norm": 2.84375,
+ "learning_rate": 5.805860805860807e-06,
+ "loss": 0.27834036350250246,
+ "mean_token_accuracy": 0.9109352678060532,
+ "num_tokens": 2709073.0,
+ "step": 290
+ },
+ {
+ "entropy": 0.2844470454379916,
+ "epoch": 0.990916597853014,
+ "grad_norm": 1.640625,
+ "learning_rate": 5.6227106227106235e-06,
+ "loss": 0.28873960971832274,
+ "mean_token_accuracy": 0.9069667845964432,
+ "num_tokens": 2805135.0,
+ "step": 300
+ },
+ {
+ "epoch": 0.990916597853014,
+ "eval_entropy": 0.285272175570329,
+ "eval_loss": 0.28705036640167236,
+ "eval_mean_token_accuracy": 0.9062359571456909,
+ "eval_num_tokens": 2805135.0,
+ "eval_runtime": 17.9038,
+ "eval_samples_per_second": 30.105,
+ "eval_steps_per_second": 15.081,
+ "step": 300
+ },
+ {
+ "entropy": 0.2869852078266633,
+ "epoch": 1.023121387283237,
+ "grad_norm": 1.9375,
+ "learning_rate": 5.43956043956044e-06,
+ "loss": 0.2850792646408081,
+ "mean_token_accuracy": 0.9064911672702203,
+ "num_tokens": 2898984.0,
+ "step": 310
+ },
+ {
+ "entropy": 0.2757460119202733,
+ "epoch": 1.0561519405450042,
+ "grad_norm": 1.8359375,
+ "learning_rate": 5.256410256410257e-06,
+ "loss": 0.2779810905456543,
+ "mean_token_accuracy": 0.9096895463764667,
+ "num_tokens": 2991694.0,
+ "step": 320
+ },
+ {
+ "entropy": 0.2600112119689584,
+ "epoch": 1.0891824938067713,
+ "grad_norm": 2.0625,
+ "learning_rate": 5.0732600732600735e-06,
+ "loss": 0.2662019729614258,
+ "mean_token_accuracy": 0.912333931028843,
+ "num_tokens": 3078862.0,
+ "step": 330
+ },
+ {
+ "entropy": 0.27342566605657337,
+ "epoch": 1.1222130470685383,
+ "grad_norm": 1.921875,
+ "learning_rate": 4.890109890109891e-06,
+ "loss": 0.28096842765808105,
+ "mean_token_accuracy": 0.9091624312102795,
+ "num_tokens": 3174256.0,
+ "step": 340
+ },
+ {
+ "entropy": 0.2671937083825469,
+ "epoch": 1.1552436003303055,
+ "grad_norm": 1.75,
+ "learning_rate": 4.706959706959707e-06,
+ "loss": 0.2723649501800537,
+ "mean_token_accuracy": 0.9101433955132961,
+ "num_tokens": 3272477.0,
+ "step": 350
+ },
+ {
+ "entropy": 0.2617026660591364,
+ "epoch": 1.1882741535920727,
+ "grad_norm": 1.9453125,
+ "learning_rate": 4.523809523809524e-06,
+ "loss": 0.26422107219696045,
+ "mean_token_accuracy": 0.913913157582283,
+ "num_tokens": 3364137.0,
+ "step": 360
+ },
+ {
+ "entropy": 0.2595691759139299,
+ "epoch": 1.2213047068538398,
+ "grad_norm": 1.5390625,
+ "learning_rate": 4.340659340659341e-06,
+ "loss": 0.2714951276779175,
+ "mean_token_accuracy": 0.911636833101511,
+ "num_tokens": 3460217.0,
+ "step": 370
+ },
+ {
+ "entropy": 0.2933320404961705,
+ "epoch": 1.254335260115607,
+ "grad_norm": 1.9765625,
+ "learning_rate": 4.157509157509158e-06,
+ "loss": 0.29573276042938235,
+ "mean_token_accuracy": 0.902192335575819,
+ "num_tokens": 3557030.0,
+ "step": 380
+ },
+ {
+ "entropy": 0.2693369995802641,
+ "epoch": 1.287365813377374,
+ "grad_norm": 1.890625,
+ "learning_rate": 3.974358974358974e-06,
+ "loss": 0.27785425186157225,
+ "mean_token_accuracy": 0.9090337291359901,
+ "num_tokens": 3654159.0,
+ "step": 390
+ },
+ {
+ "entropy": 0.2908114368095994,
+ "epoch": 1.3203963666391412,
+ "grad_norm": 2.015625,
+ "learning_rate": 3.7912087912087915e-06,
+ "loss": 0.30535905361175536,
+ "mean_token_accuracy": 0.901054035872221,
+ "num_tokens": 3749664.0,
+ "step": 400
+ },
+ {
+ "epoch": 1.3203963666391412,
+ "eval_entropy": 0.2783401116176888,
+ "eval_loss": 0.28495287895202637,
+ "eval_mean_token_accuracy": 0.9069356251645971,
+ "eval_num_tokens": 3749664.0,
+ "eval_runtime": 17.8814,
+ "eval_samples_per_second": 30.143,
+ "eval_steps_per_second": 15.099,
+ "step": 400
+ },
+ {
+ "entropy": 0.2749379588291049,
+ "epoch": 1.3534269199009084,
+ "grad_norm": 1.84375,
+ "learning_rate": 3.6080586080586083e-06,
+ "loss": 0.27795455455780027,
+ "mean_token_accuracy": 0.9090839587152004,
+ "num_tokens": 3845841.0,
+ "step": 410
+ },
+ {
+ "entropy": 0.2859474029392004,
+ "epoch": 1.3864574731626755,
+ "grad_norm": 1.6796875,
+ "learning_rate": 3.424908424908425e-06,
+ "loss": 0.2886682987213135,
+ "mean_token_accuracy": 0.9059950634837151,
+ "num_tokens": 3942402.0,
+ "step": 420
+ },
+ {
+ "entropy": 0.2537162031978369,
+ "epoch": 1.4194880264244425,
+ "grad_norm": 1.9375,
+ "learning_rate": 3.2417582417582424e-06,
+ "loss": 0.25533475875854494,
+ "mean_token_accuracy": 0.9158934533596039,
+ "num_tokens": 4034122.0,
+ "step": 430
+ },
+ {
+ "entropy": 0.28255065642297267,
+ "epoch": 1.4525185796862097,
+ "grad_norm": 1.7734375,
+ "learning_rate": 3.058608058608059e-06,
+ "loss": 0.29817631244659426,
+ "mean_token_accuracy": 0.9040707617998123,
+ "num_tokens": 4127068.0,
+ "step": 440
+ },
+ {
+ "entropy": 0.2898306529968977,
+ "epoch": 1.4855491329479769,
+ "grad_norm": 1.703125,
+ "learning_rate": 2.8754578754578755e-06,
+ "loss": 0.2923931121826172,
+ "mean_token_accuracy": 0.9030195340514183,
+ "num_tokens": 4224435.0,
+ "step": 450
+ },
+ {
+ "entropy": 0.27209411058574917,
+ "epoch": 1.518579686209744,
+ "grad_norm": 1.7578125,
+ "learning_rate": 2.6923076923076923e-06,
+ "loss": 0.28646528720855713,
+ "mean_token_accuracy": 0.9095881894230843,
+ "num_tokens": 4321028.0,
+ "step": 460
+ },
+ {
+ "entropy": 0.280727200768888,
+ "epoch": 1.5516102394715112,
+ "grad_norm": 1.7421875,
+ "learning_rate": 2.509157509157509e-06,
+ "loss": 0.2891413688659668,
+ "mean_token_accuracy": 0.9067636162042618,
+ "num_tokens": 4414916.0,
+ "step": 470
+ },
+ {
+ "entropy": 0.27109829243272543,
+ "epoch": 1.5846407927332784,
+ "grad_norm": 1.8359375,
+ "learning_rate": 2.3260073260073264e-06,
+ "loss": 0.26625516414642336,
+ "mean_token_accuracy": 0.9099020563066006,
+ "num_tokens": 4505708.0,
+ "step": 480
+ },
+ {
+ "entropy": 0.2729395367205143,
+ "epoch": 1.6176713459950454,
+ "grad_norm": 1.7890625,
+ "learning_rate": 2.1428571428571427e-06,
+ "loss": 0.2779061317443848,
+ "mean_token_accuracy": 0.909747340530157,
+ "num_tokens": 4600244.0,
+ "step": 490
+ },
+ {
+ "entropy": 0.2682301852852106,
+ "epoch": 1.6507018992568125,
+ "grad_norm": 1.6953125,
+ "learning_rate": 1.95970695970696e-06,
+ "loss": 0.274446964263916,
+ "mean_token_accuracy": 0.9094639919698239,
+ "num_tokens": 4688147.0,
+ "step": 500
+ },
+ {
+ "epoch": 1.6507018992568125,
+ "eval_entropy": 0.2756172451708052,
+ "eval_loss": 0.2841346263885498,
+ "eval_mean_token_accuracy": 0.9072470570052111,
+ "eval_num_tokens": 4688147.0,
+ "eval_runtime": 17.9119,
+ "eval_samples_per_second": 30.092,
+ "eval_steps_per_second": 15.074,
+ "step": 500
+ },
+ {
+ "entropy": 0.26342453975230457,
+ "epoch": 1.6837324525185797,
+ "grad_norm": 1.890625,
+ "learning_rate": 1.7765567765567768e-06,
+ "loss": 0.26432271003723146,
+ "mean_token_accuracy": 0.9114409819245338,
+ "num_tokens": 4780124.0,
+ "step": 510
+ },
+ {
+ "entropy": 0.28657868495211003,
+ "epoch": 1.7167630057803467,
+ "grad_norm": 1.7734375,
+ "learning_rate": 1.5934065934065933e-06,
+ "loss": 0.2983593225479126,
+ "mean_token_accuracy": 0.9028328903019428,
+ "num_tokens": 4874633.0,
+ "step": 520
+ },
+ {
+ "entropy": 0.25633321776986123,
+ "epoch": 1.7497935590421139,
+ "grad_norm": 1.8359375,
+ "learning_rate": 1.4102564102564104e-06,
+ "loss": 0.2606488227844238,
+ "mean_token_accuracy": 0.9153999648988247,
+ "num_tokens": 4967263.0,
+ "step": 530
+ },
+ {
+ "entropy": 0.26120030507445335,
+ "epoch": 1.782824112303881,
+ "grad_norm": 2.078125,
+ "learning_rate": 1.2271062271062271e-06,
+ "loss": 0.2706313610076904,
+ "mean_token_accuracy": 0.9130682416260243,
+ "num_tokens": 5056736.0,
+ "step": 540
+ },
+ {
+ "entropy": 0.27009565513581035,
+ "epoch": 1.8158546655656482,
+ "grad_norm": 1.8828125,
+ "learning_rate": 1.0439560439560442e-06,
+ "loss": 0.277858567237854,
+ "mean_token_accuracy": 0.9087999157607556,
+ "num_tokens": 5150617.0,
+ "step": 550
+ },
+ {
+ "entropy": 0.2819518899545074,
+ "epoch": 1.8488852188274154,
+ "grad_norm": 1.9609375,
+ "learning_rate": 8.608058608058609e-07,
+ "loss": 0.28743035793304444,
+ "mean_token_accuracy": 0.9067244723439216,
+ "num_tokens": 5243538.0,
+ "step": 560
+ },
+ {
+ "entropy": 0.2858045116066933,
+ "epoch": 1.8819157720891826,
+ "grad_norm": 2.203125,
+ "learning_rate": 6.776556776556777e-07,
+ "loss": 0.2939825296401978,
+ "mean_token_accuracy": 0.904579871147871,
+ "num_tokens": 5331441.0,
+ "step": 570
+ },
+ {
+ "entropy": 0.2926233192905784,
+ "epoch": 1.9149463253509498,
+ "grad_norm": 2.3125,
+ "learning_rate": 4.945054945054946e-07,
+ "loss": 0.29961700439453126,
+ "mean_token_accuracy": 0.9033655568957328,
+ "num_tokens": 5426734.0,
+ "step": 580
+ },
+ {
+ "entropy": 0.2706678254529834,
+ "epoch": 1.9479768786127167,
+ "grad_norm": 2.03125,
+ "learning_rate": 3.113553113553114e-07,
+ "loss": 0.27953526973724363,
+ "mean_token_accuracy": 0.9093762949109078,
+ "num_tokens": 5518190.0,
+ "step": 590
+ },
+ {
+ "entropy": 0.27056904807686805,
+ "epoch": 1.981007431874484,
+ "grad_norm": 2.09375,
+ "learning_rate": 1.282051282051282e-07,
+ "loss": 0.2764536619186401,
+ "mean_token_accuracy": 0.9110228553414345,
+ "num_tokens": 5609238.0,
+ "step": 600
+ },
+ {
+ "epoch": 1.981007431874484,
+ "eval_entropy": 0.2757842331572815,
+ "eval_loss": 0.28394654393196106,
+ "eval_mean_token_accuracy": 0.9073928307603907,
+ "eval_num_tokens": 5609238.0,
+ "eval_runtime": 18.1138,
+ "eval_samples_per_second": 29.756,
+ "eval_steps_per_second": 14.906,
+ "step": 600
+ },
+ {
+ "epoch": 2.0,
+ "eval_entropy": 0.2757109238593667,
+ "eval_loss": 0.2840007245540619,
+ "eval_mean_token_accuracy": 0.9072579809912928,
+ "eval_num_tokens": 5663138.0,
+ "eval_runtime": 18.1948,
+ "eval_samples_per_second": 29.624,
+ "eval_steps_per_second": 14.839,
+ "step": 606
+ }
+ ],
+ "logging_steps": 10,
+ "max_steps": 606,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 2,
+ "save_steps": 100,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 6.072645019326874e+16,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-606/training_args.bin b/checkpoint-606/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..133cb6661b0124bb9db0d7d208d748fd736ea689
--- /dev/null
+++ b/checkpoint-606/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6696e1fef8bde5e22de2f5ae18c284f2b2d35a345157db5151571518385fb38b
+size 5713
diff --git a/checkpoint-674/chat_template.jinja b/checkpoint-674/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..01be9b307daa2d425f7c168c9fb145a286e0afb4
--- /dev/null
+++ b/checkpoint-674/chat_template.jinja
@@ -0,0 +1,89 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0].role == 'system' %}
+ {{- messages[0].content + '\n\n' }}
+ {%- endif %}
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0].role == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+ {%- set index = (messages|length - 1) - loop.index0 %}
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}
+ {%- set ns.multi_step_tool = false %}
+ {%- set ns.last_query_index = index %}
+ {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+ {%- if message.content is string %}
+ {%- set content = message.content %}
+ {%- else %}
+ {%- set content = '' %}
+ {%- endif %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {%- set reasoning_content = '' %}
+ {%- if message.reasoning_content is string %}
+ {%- set reasoning_content = message.reasoning_content %}
+ {%- else %}
+ {%- if '' in content %}
+ {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %}
+ {%- set content = content.split('')[-1].lstrip('\n') %}
+ {%- endif %}
+ {%- endif %}
+ {%- if loop.index0 > ns.last_query_index %}
+ {%- if loop.last or (not loop.last and reasoning_content) %}
+ {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- else %}
+ {{- '<|im_start|>' + message.role + '\n' + content }}
+ {%- endif %}
+ {%- if message.tool_calls %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if (loop.first and content) or (not loop.first) %}
+ {{- '\n' }}
+ {%- endif %}
+ {%- if tool_call.function %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {%- if tool_call.arguments is string %}
+ {{- tool_call.arguments }}
+ {%- else %}
+ {{- tool_call.arguments | tojson }}
+ {%- endif %}
+ {{- '}\n' }}
+ {%- endfor %}
+ {%- endif %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+ {%- if enable_thinking is defined and enable_thinking is false %}
+ {{- '\n\n\n\n' }}
+ {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/checkpoint-674/config.json b/checkpoint-674/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d219d6a6d288625229a71646710b85cf7ae5def
--- /dev/null
+++ b/checkpoint-674/config.json
@@ -0,0 +1,63 @@
+{
+ "architectures": [
+ "Qwen3ForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "bfloat16",
+ "eos_token_id": 151645,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 6144,
+ "layer_types": [
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 40960,
+ "max_window_layers": 28,
+ "model_type": "qwen3",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 8,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.7.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/checkpoint-674/generation_config.json b/checkpoint-674/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec133d867b1427d43a5ee0ea332bd088fc12013
--- /dev/null
+++ b/checkpoint-674/generation_config.json
@@ -0,0 +1,12 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "temperature": 0.6,
+ "top_k": 20,
+ "top_p": 0.95,
+ "transformers_version": "5.7.0"
+}
diff --git a/checkpoint-674/model.safetensors b/checkpoint-674/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cc885a501a0fffef97b843b5248e3bb7cfe46172
--- /dev/null
+++ b/checkpoint-674/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3af4557502ae7c7853ed3b684aa50cfd00317aafc237f2dda688075bb3dc81c1
+size 3441185608
diff --git a/checkpoint-674/optimizer.pt b/checkpoint-674/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4f8c22d5c745517a99843c614c05da8a366a7d78
--- /dev/null
+++ b/checkpoint-674/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ca0ddbeff189519bd91d1474687c97eb42f33e1dc183a5769974dadc5e0cd7b
+size 6882572207
diff --git a/checkpoint-674/rng_state.pth b/checkpoint-674/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..1feba1a6538e93b94696d3773853dbc8947b0cad
--- /dev/null
+++ b/checkpoint-674/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:098b29492211804ab324a36f37466821d948280bb74fce4ba895c03f13ecd878
+size 14645
diff --git a/checkpoint-674/scheduler.pt b/checkpoint-674/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..52e04de5969d9a926e44ed315d79199b8ca27698
--- /dev/null
+++ b/checkpoint-674/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01dbf241e935857d2a5486058c590aafb795b79d6ef3f0dc038c0d85e2187d7d
+size 1465
diff --git a/checkpoint-674/tokenizer.json b/checkpoint-674/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7afbed2efcdf019f88ab0572ec29d3bf595dfe2
--- /dev/null
+++ b/checkpoint-674/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650
diff --git a/checkpoint-674/tokenizer_config.json b/checkpoint-674/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/checkpoint-674/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "local_files_only": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/checkpoint-674/trainer_state.json b/checkpoint-674/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..54b5d973ad7d8fb6a649dbc5f72a0c5653d31516
--- /dev/null
+++ b/checkpoint-674/trainer_state.json
@@ -0,0 +1,704 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 2.0,
+ "eval_steps": 500,
+ "global_step": 674,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "entropy": NaN,
+ "epoch": 0.029717682020802376,
+ "grad_norm": NaN,
+ "learning_rate": 1.8000000000000001e-06,
+ "loss": 1005514863411.2,
+ "mean_token_accuracy": 0.07770282882993343,
+ "num_tokens": 97764.0,
+ "step": 10
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.05943536404160475,
+ "grad_norm": NaN,
+ "learning_rate": 3.8000000000000005e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007392044077278115,
+ "num_tokens": 190372.0,
+ "step": 20
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.08915304606240713,
+ "grad_norm": NaN,
+ "learning_rate": 5.8e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000383764869911829,
+ "num_tokens": 280928.0,
+ "step": 30
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.1188707280832095,
+ "grad_norm": NaN,
+ "learning_rate": 7.800000000000002e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003212536917999387,
+ "num_tokens": 380644.0,
+ "step": 40
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.1485884101040119,
+ "grad_norm": NaN,
+ "learning_rate": 9.800000000000001e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003853117574180942,
+ "num_tokens": 475467.0,
+ "step": 50
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.17830609212481427,
+ "grad_norm": NaN,
+ "learning_rate": 1.18e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003134153150313068,
+ "num_tokens": 572411.0,
+ "step": 60
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.20802377414561665,
+ "grad_norm": NaN,
+ "learning_rate": 1.38e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011001345737895463,
+ "num_tokens": 661517.0,
+ "step": 70
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.237741456166419,
+ "grad_norm": NaN,
+ "learning_rate": 1.58e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012231820757733658,
+ "num_tokens": 752142.0,
+ "step": 80
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.2674591381872214,
+ "grad_norm": NaN,
+ "learning_rate": 1.7800000000000002e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011551132018212229,
+ "num_tokens": 850399.0,
+ "step": 90
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.2971768202080238,
+ "grad_norm": NaN,
+ "learning_rate": 1.98e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00047943509634933433,
+ "num_tokens": 954759.0,
+ "step": 100
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.32689450222882616,
+ "grad_norm": NaN,
+ "learning_rate": 1.9802414928649837e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008438902725174558,
+ "num_tokens": 1050439.0,
+ "step": 110
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.35661218424962854,
+ "grad_norm": NaN,
+ "learning_rate": 1.958287596048299e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009903057703922967,
+ "num_tokens": 1135089.0,
+ "step": 120
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.3863298662704309,
+ "grad_norm": NaN,
+ "learning_rate": 1.9363336992316137e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00040032515971688554,
+ "num_tokens": 1228149.0,
+ "step": 130
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.4160475482912333,
+ "grad_norm": NaN,
+ "learning_rate": 1.9143798024149288e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0010247966441966128,
+ "num_tokens": 1331672.0,
+ "step": 140
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.4457652303120357,
+ "grad_norm": NaN,
+ "learning_rate": 1.892425905598244e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000843161555530969,
+ "num_tokens": 1421314.0,
+ "step": 150
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.475482912332838,
+ "grad_norm": NaN,
+ "learning_rate": 1.8704720087815587e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007889680869993754,
+ "num_tokens": 1511701.0,
+ "step": 160
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5052005943536404,
+ "grad_norm": NaN,
+ "learning_rate": 1.848518111964874e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007493881763366517,
+ "num_tokens": 1605063.0,
+ "step": 170
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5349182763744428,
+ "grad_norm": NaN,
+ "learning_rate": 1.826564215148189e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000567350817436818,
+ "num_tokens": 1703107.0,
+ "step": 180
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5646359583952452,
+ "grad_norm": NaN,
+ "learning_rate": 1.804610318331504e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00027060438151238486,
+ "num_tokens": 1796279.0,
+ "step": 190
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.5943536404160475,
+ "grad_norm": NaN,
+ "learning_rate": 1.782656421514819e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008226798396208324,
+ "num_tokens": 1893647.0,
+ "step": 200
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.6240713224368499,
+ "grad_norm": NaN,
+ "learning_rate": 1.760702524698134e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007338574505411088,
+ "num_tokens": 1982325.0,
+ "step": 210
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.6537890044576523,
+ "grad_norm": NaN,
+ "learning_rate": 1.7387486278814492e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004891513846814633,
+ "num_tokens": 2071869.0,
+ "step": 220
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.6835066864784547,
+ "grad_norm": NaN,
+ "learning_rate": 1.716794731064764e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00038509477453771976,
+ "num_tokens": 2168370.0,
+ "step": 230
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.7132243684992571,
+ "grad_norm": NaN,
+ "learning_rate": 1.694840834248079e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009255842014681548,
+ "num_tokens": 2263211.0,
+ "step": 240
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.7429420505200595,
+ "grad_norm": NaN,
+ "learning_rate": 1.6728869374313943e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011857455348945223,
+ "num_tokens": 2352425.0,
+ "step": 250
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.7726597325408618,
+ "grad_norm": NaN,
+ "learning_rate": 1.6509330406147094e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008204675337765366,
+ "num_tokens": 2445217.0,
+ "step": 260
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.8023774145616642,
+ "grad_norm": NaN,
+ "learning_rate": 1.6289791437980242e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0013553448639868292,
+ "num_tokens": 2537677.0,
+ "step": 270
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.8320950965824666,
+ "grad_norm": NaN,
+ "learning_rate": 1.6070252469813393e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011630529144895263,
+ "num_tokens": 2631646.0,
+ "step": 280
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.861812778603269,
+ "grad_norm": NaN,
+ "learning_rate": 1.5850713501646545e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009409735575900413,
+ "num_tokens": 2722632.0,
+ "step": 290
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.8915304606240714,
+ "grad_norm": NaN,
+ "learning_rate": 1.5631174533479693e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006484775040007662,
+ "num_tokens": 2819185.0,
+ "step": 300
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.9212481426448736,
+ "grad_norm": NaN,
+ "learning_rate": 1.5411635565312844e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011682231881422922,
+ "num_tokens": 2910483.0,
+ "step": 310
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.950965824665676,
+ "grad_norm": NaN,
+ "learning_rate": 1.5192096597145995e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004708930806373246,
+ "num_tokens": 2997057.0,
+ "step": 320
+ },
+ {
+ "entropy": NaN,
+ "epoch": 0.9806835066864784,
+ "grad_norm": NaN,
+ "learning_rate": 1.4972557628979145e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011955970534472727,
+ "num_tokens": 3091284.0,
+ "step": 330
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.0089153046062407,
+ "grad_norm": NaN,
+ "learning_rate": 1.4753018660812295e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006843029964090276,
+ "num_tokens": 3177800.0,
+ "step": 340
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.0386329866270432,
+ "grad_norm": NaN,
+ "learning_rate": 1.4533479692645444e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0005406756106822285,
+ "num_tokens": 3271989.0,
+ "step": 350
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.0683506686478454,
+ "grad_norm": NaN,
+ "learning_rate": 1.4313940724478596e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007009803419350646,
+ "num_tokens": 3365082.0,
+ "step": 360
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.098068350668648,
+ "grad_norm": NaN,
+ "learning_rate": 1.4094401756311746e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001003894711175235,
+ "num_tokens": 3465865.0,
+ "step": 370
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.1277860326894502,
+ "grad_norm": NaN,
+ "learning_rate": 1.3874862788144897e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004136150790145621,
+ "num_tokens": 3564998.0,
+ "step": 380
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.1575037147102525,
+ "grad_norm": NaN,
+ "learning_rate": 1.3655323819978048e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000998112405068241,
+ "num_tokens": 3661508.0,
+ "step": 390
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.187221396731055,
+ "grad_norm": NaN,
+ "learning_rate": 1.3435784851811198e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009621107514249161,
+ "num_tokens": 3755395.0,
+ "step": 400
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.2169390787518575,
+ "grad_norm": NaN,
+ "learning_rate": 1.3216245883644348e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004976072828867473,
+ "num_tokens": 3842994.0,
+ "step": 410
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.2466567607726597,
+ "grad_norm": NaN,
+ "learning_rate": 1.2996706915477499e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0003434549740632065,
+ "num_tokens": 3934618.0,
+ "step": 420
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.276374442793462,
+ "grad_norm": NaN,
+ "learning_rate": 1.2777167947310649e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00026364591685705817,
+ "num_tokens": 4026711.0,
+ "step": 430
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3060921248142645,
+ "grad_norm": NaN,
+ "learning_rate": 1.2557628979143798e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006349188319290988,
+ "num_tokens": 4114609.0,
+ "step": 440
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3358098068350668,
+ "grad_norm": NaN,
+ "learning_rate": 1.2338090010976948e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004265022376785055,
+ "num_tokens": 4203953.0,
+ "step": 450
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3655274888558693,
+ "grad_norm": NaN,
+ "learning_rate": 1.2118551042810101e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012633154692593962,
+ "num_tokens": 4298319.0,
+ "step": 460
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.3952451708766715,
+ "grad_norm": NaN,
+ "learning_rate": 1.189901207464325e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0011256658195634372,
+ "num_tokens": 4396020.0,
+ "step": 470
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.424962852897474,
+ "grad_norm": NaN,
+ "learning_rate": 1.16794731064764e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0013131745137798135,
+ "num_tokens": 4491203.0,
+ "step": 480
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.4546805349182763,
+ "grad_norm": NaN,
+ "learning_rate": 1.1459934138309552e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012716218378045596,
+ "num_tokens": 4582199.0,
+ "step": 490
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.4843982169390788,
+ "grad_norm": NaN,
+ "learning_rate": 1.1240395170142701e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0014869970778818241,
+ "num_tokens": 4672589.0,
+ "step": 500
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.514115898959881,
+ "grad_norm": NaN,
+ "learning_rate": 1.1020856201975851e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00046981508567114363,
+ "num_tokens": 4767693.0,
+ "step": 510
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.5438335809806834,
+ "grad_norm": NaN,
+ "learning_rate": 1.0801317233809e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000601445968641201,
+ "num_tokens": 4860371.0,
+ "step": 520
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.5735512630014858,
+ "grad_norm": NaN,
+ "learning_rate": 1.0581778265642154e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012073652920662425,
+ "num_tokens": 4950622.0,
+ "step": 530
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6032689450222883,
+ "grad_norm": NaN,
+ "learning_rate": 1.0362239297475303e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.00028707647215924225,
+ "num_tokens": 5044878.0,
+ "step": 540
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6329866270430906,
+ "grad_norm": NaN,
+ "learning_rate": 1.0142700329308453e-05,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004101752907445189,
+ "num_tokens": 5140938.0,
+ "step": 550
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6627043090638929,
+ "grad_norm": NaN,
+ "learning_rate": 9.923161361141603e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008072987395280507,
+ "num_tokens": 5232852.0,
+ "step": 560
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.6924219910846954,
+ "grad_norm": NaN,
+ "learning_rate": 9.703622392974754e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0012222853780258446,
+ "num_tokens": 5321835.0,
+ "step": 570
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.7221396731054979,
+ "grad_norm": NaN,
+ "learning_rate": 9.484083424807904e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001064640874392353,
+ "num_tokens": 5416464.0,
+ "step": 580
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.7518573551263001,
+ "grad_norm": NaN,
+ "learning_rate": 9.264544456641055e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0009612867230316624,
+ "num_tokens": 5515169.0,
+ "step": 590
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.7815750371471024,
+ "grad_norm": NaN,
+ "learning_rate": 9.045005488474205e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008774219051701948,
+ "num_tokens": 5611971.0,
+ "step": 600
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.811292719167905,
+ "grad_norm": NaN,
+ "learning_rate": 8.825466520307355e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007863394406740553,
+ "num_tokens": 5704255.0,
+ "step": 610
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.8410104011887074,
+ "grad_norm": NaN,
+ "learning_rate": 8.605927552140506e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.000578320066415472,
+ "num_tokens": 5800440.0,
+ "step": 620
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.8707280832095097,
+ "grad_norm": NaN,
+ "learning_rate": 8.386388583973656e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0004275172817870043,
+ "num_tokens": 5890898.0,
+ "step": 630
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.900445765230312,
+ "grad_norm": NaN,
+ "learning_rate": 8.166849615806807e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.001425291335908696,
+ "num_tokens": 5987228.0,
+ "step": 640
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.9301634472511144,
+ "grad_norm": NaN,
+ "learning_rate": 7.947310647639957e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0006059414023184218,
+ "num_tokens": 6076579.0,
+ "step": 650
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.959881129271917,
+ "grad_norm": NaN,
+ "learning_rate": 7.727771679473108e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0008722793892957271,
+ "num_tokens": 6171013.0,
+ "step": 660
+ },
+ {
+ "entropy": NaN,
+ "epoch": 1.9895988112927192,
+ "grad_norm": NaN,
+ "learning_rate": 7.508232711306257e-06,
+ "loss": 0.0,
+ "mean_token_accuracy": 0.0007748565993097145,
+ "num_tokens": 6265432.0,
+ "step": 670
+ }
+ ],
+ "logging_steps": 10,
+ "max_steps": 1011,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 6.753063653143757e+16,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-674/training_args.bin b/checkpoint-674/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..737a87caa4fc2f704a1fe56a508ca4d335ca783d
--- /dev/null
+++ b/checkpoint-674/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae0f7084660199ed7fd16027c0c6af5e726790f32a022b019f1277bec4e01142
+size 5713
diff --git a/config.json b/config.json
index cd2e0cfcc25e9201303ab4390eee904c9cc5a285..c849622762e06273c720e07098ab07dbbea3b8b0 100644
--- a/config.json
+++ b/config.json
@@ -5,7 +5,7 @@
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": null,
- "dtype": "bfloat16",
+ "dtype": "float32",
"eos_token_id": 151645,
"head_dim": 128,
"hidden_act": "silu",
diff --git a/eval_results.diverse.json b/eval_results.diverse.json
new file mode 100644
index 0000000000000000000000000000000000000000..1f6ec55ddc2f6134480996e20074310f889d6b16
--- /dev/null
+++ b/eval_results.diverse.json
@@ -0,0 +1,207 @@
+{
+ "checkpoint": "/scratch/checkpoints/exp6_sft_numinamath_dpo",
+ "is_local": true,
+ "eval_tier": "diverse",
+ "sampling": {
+ "n": 8,
+ "temperature": 0.6,
+ "top_p": 0.95,
+ "top_k": 20,
+ "max_tokens": 3500
+ },
+ "benchmarks": {
+ "numinamath_benchmark": {
+ "source": "numinamath_benchmark",
+ "overall": {
+ "n_problems": 500,
+ "n_completions": 8,
+ "pass@1": 0.3795,
+ "pass@8": 0.612
+ },
+ "by_source": {
+ "numina:amc_aime": {
+ "n_problems": 1,
+ "n_completions": 8,
+ "pass@1": 0.25,
+ "pass@8": 1.0
+ },
+ "numina:aops_forum": {
+ "n_problems": 9,
+ "n_completions": 8,
+ "pass@1": 0.08333333333333333,
+ "pass@8": 0.4444444444444444
+ },
+ "numina:cn_contest": {
+ "n_problems": 7,
+ "n_completions": 8,
+ "pass@1": 0.0,
+ "pass@8": 0.0
+ },
+ "numina:cn_k12": {
+ "n_problems": 200,
+ "n_completions": 8,
+ "pass@1": 0.331875,
+ "pass@8": 0.62
+ },
+ "numina:inequalities": {
+ "n_problems": 1,
+ "n_completions": 8,
+ "pass@1": 0.0,
+ "pass@8": 0.0
+ },
+ "numina:metamath": {
+ "n_problems": 7,
+ "n_completions": 8,
+ "pass@1": 0.1964285714285714,
+ "pass@8": 0.7142857142857143
+ },
+ "numina:olympiads": {
+ "n_problems": 49,
+ "n_completions": 8,
+ "pass@1": 0.06887755102040816,
+ "pass@8": 0.20408163265306123
+ },
+ "numina:olympiads_ref": {
+ "n_problems": 1,
+ "n_completions": 8,
+ "pass@1": 0.25,
+ "pass@8": 1.0
+ },
+ "numina:orca_math": {
+ "n_problems": 112,
+ "n_completions": 8,
+ "pass@1": 0.6149553571428571,
+ "pass@8": 0.8125
+ },
+ "numina:synthetic_math": {
+ "n_problems": 113,
+ "n_completions": 8,
+ "pass@1": 0.42920353982300885,
+ "pass@8": 0.6194690265486725
+ }
+ },
+ "by_topic": {
+ "Algebra": {
+ "n_problems": 277,
+ "n_completions": 8,
+ "pass@1": 0.4453971119133574,
+ "pass@8": 0.6895306859205776
+ },
+ "Calculus": {
+ "n_problems": 17,
+ "n_completions": 8,
+ "pass@1": 0.20588235294117646,
+ "pass@8": 0.35294117647058826
+ },
+ "Combinatorics": {
+ "n_problems": 31,
+ "n_completions": 8,
+ "pass@1": 0.2862903225806452,
+ "pass@8": 0.5806451612903226
+ },
+ "Geometry": {
+ "n_problems": 84,
+ "n_completions": 8,
+ "pass@1": 0.3169642857142857,
+ "pass@8": 0.5357142857142857
+ },
+ "Inequalities": {
+ "n_problems": 11,
+ "n_completions": 8,
+ "pass@1": 0.375,
+ "pass@8": 0.6363636363636364
+ },
+ "Logic and Puzzles": {
+ "n_problems": 20,
+ "n_completions": 8,
+ "pass@1": 0.45,
+ "pass@8": 0.65
+ },
+ "Number Theory": {
+ "n_problems": 54,
+ "n_completions": 8,
+ "pass@1": 0.24537037037037038,
+ "pass@8": 0.46296296296296297
+ },
+ "Other": {
+ "n_problems": 6,
+ "n_completions": 8,
+ "pass@1": 0.16666666666666666,
+ "pass@8": 0.16666666666666666
+ }
+ }
+ },
+ "diverse_math_benchmark": {
+ "source": "diverse_math_benchmark",
+ "overall": {
+ "n_problems": 150,
+ "n_completions": 8,
+ "pass@1": 0.3275,
+ "pass@8": 0.5466666666666666
+ },
+ "by_source": {
+ "aime_2024": {
+ "n_problems": 17,
+ "n_completions": 8,
+ "pass@1": 0.036764705882352935,
+ "pass@8": 0.11764705882352941
+ },
+ "aime_2025": {
+ "n_problems": 13,
+ "n_completions": 8,
+ "pass@1": 0.08653846153846154,
+ "pass@8": 0.15384615384615385
+ },
+ "math_hard": {
+ "n_problems": 50,
+ "n_completions": 8,
+ "pass@1": 0.495,
+ "pass@8": 0.78
+ },
+ "numina:amc_aime": {
+ "n_problems": 1,
+ "n_completions": 8,
+ "pass@1": 0.125,
+ "pass@8": 1.0
+ },
+ "numina:aops_forum": {
+ "n_problems": 2,
+ "n_completions": 8,
+ "pass@1": 0.0,
+ "pass@8": 0.0
+ },
+ "numina:cn_k12": {
+ "n_problems": 26,
+ "n_completions": 8,
+ "pass@1": 0.32211538461538464,
+ "pass@8": 0.5769230769230769
+ },
+ "numina:metamath": {
+ "n_problems": 1,
+ "n_completions": 8,
+ "pass@1": 0.0,
+ "pass@8": 0.0
+ },
+ "numina:olympiads": {
+ "n_problems": 17,
+ "n_completions": 8,
+ "pass@1": 0.08088235294117647,
+ "pass@8": 0.29411764705882354
+ },
+ "numina:orca_math": {
+ "n_problems": 14,
+ "n_completions": 8,
+ "pass@1": 0.6964285714285714,
+ "pass@8": 0.9285714285714286
+ },
+ "numina:synthetic_math": {
+ "n_problems": 9,
+ "n_completions": 8,
+ "pass@1": 0.3333333333333333,
+ "pass@8": 0.5555555555555556
+ }
+ }
+ }
+ },
+ "note": "All benchmarks use identical sampling params so numbers compare directly across experiments (baseline, SFT v1/v2, GRPO). Hard benchmarks (math_hard_100, aime_benchmark) are decontaminated against MixLong (the SFT pool)."
+}
\ No newline at end of file
diff --git a/eval_results.json b/eval_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..34bf083a93ff87791f872d45ce86fa8c43f28737
--- /dev/null
+++ b/eval_results.json
@@ -0,0 +1,114 @@
+{
+ "checkpoint": "/scratch/checkpoints/exp6_sft_numinamath_dpo",
+ "is_local": true,
+ "eval_tier": "fast",
+ "sampling": {
+ "n": 8,
+ "temperature": 0.3,
+ "top_p": 0.95,
+ "top_k": 20,
+ "max_tokens": 3500
+ },
+ "benchmarks": {
+ "ci_gate_math": {
+ "source": "ci_gate_math",
+ "overall": {
+ "n_problems": 10,
+ "n_completions": 8,
+ "pass@1": 0.2,
+ "pass@8": 0.3
+ }
+ },
+ "math_hard_100": {
+ "source": "math_hard_100",
+ "overall": {
+ "n_problems": 100,
+ "n_completions": 8,
+ "pass@1": 0.49375,
+ "pass@8": 0.75
+ },
+ "by_level": {
+ "4": {
+ "n_problems": 51,
+ "n_completions": 8,
+ "pass@1": 0.6029411764705882,
+ "pass@8": 0.8431372549019608
+ },
+ "5": {
+ "n_problems": 49,
+ "n_completions": 8,
+ "pass@1": 0.38010204081632654,
+ "pass@8": 0.6530612244897959
+ }
+ },
+ "by_topic": {
+ "algebra": {
+ "n_problems": 15,
+ "n_completions": 8,
+ "pass@1": 0.7916666666666666,
+ "pass@8": 0.9333333333333333
+ },
+ "counting_and_probability": {
+ "n_problems": 14,
+ "n_completions": 8,
+ "pass@1": 0.2857142857142857,
+ "pass@8": 0.5
+ },
+ "geometry": {
+ "n_problems": 14,
+ "n_completions": 8,
+ "pass@1": 0.3928571428571428,
+ "pass@8": 0.7142857142857143
+ },
+ "intermediate_algebra": {
+ "n_problems": 15,
+ "n_completions": 8,
+ "pass@1": 0.4166666666666667,
+ "pass@8": 0.6666666666666666
+ },
+ "number_theory": {
+ "n_problems": 14,
+ "n_completions": 8,
+ "pass@1": 0.5178571428571429,
+ "pass@8": 0.7857142857142857
+ },
+ "prealgebra": {
+ "n_problems": 14,
+ "n_completions": 8,
+ "pass@1": 0.6607142857142857,
+ "pass@8": 1.0
+ },
+ "precalculus": {
+ "n_problems": 14,
+ "n_completions": 8,
+ "pass@1": 0.375,
+ "pass@8": 0.6428571428571429
+ }
+ }
+ },
+ "aime_benchmark": {
+ "source": "aime_benchmark",
+ "overall": {
+ "n_problems": 60,
+ "n_completions": 8,
+ "pass@1": 0.031249999999999993,
+ "pass@8": 0.11666666666666667
+ },
+ "by_source": {
+ "aime_2024": {
+ "n_problems": 30,
+ "n_completions": 8,
+ "pass@1": 0.029166666666666664,
+ "pass@8": 0.13333333333333333
+ },
+ "aime_2025": {
+ "n_problems": 30,
+ "n_completions": 8,
+ "pass@1": 0.033333333333333326,
+ "pass@8": 0.1
+ }
+ }
+ }
+ },
+ "note": "All benchmarks use identical sampling params so numbers compare directly across experiments (baseline, SFT v1/v2, GRPO). Hard benchmarks (math_hard_100, aime_benchmark) are decontaminated against MixLong (the SFT pool)."
+}
\ No newline at end of file
diff --git a/generation_config.json b/generation_config.json
index 9526fbb9cfca6c6b8911fa4f1995a5c43139fc63..ad067e1fdcd8ba85ed8a022901e87bef92fc090b 100644
--- a/generation_config.json
+++ b/generation_config.json
@@ -1,5 +1,4 @@
{
- "bos_token_id": 151643,
"do_sample": true,
"eos_token_id": [
151645,
@@ -7,8 +6,8 @@
],
"pad_token_id": 151643,
"temperature": 0.7,
- "top_p": 0.95,
"top_k": 20,
- "repetition_penalty": 1.03,
- "transformers_version": "4.51.0"
+ "top_p": 0.95,
+ "transformers_version": "5.8.0",
+ "bos_token_id": 151643
}
\ No newline at end of file
diff --git a/model.safetensors b/model.safetensors
index a9923328aed6324f3b8bc4b96ecf4cc43763ec69..2d01a55a8c76a46ec1d94126ad94dfe8a1bc4c29 100644
--- a/model.safetensors
+++ b/model.safetensors
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:c9179fe220727db304cb809795bd582e6dbc90a763ef3579708273b8b09858e5
-size 3441185608
+oid sha256:23c085561ee4f5825163d83ad9fc5cc7accfaac86a241a96d54426232a45e20e
+size 6882335328
diff --git a/split_info.json b/split_info.json
new file mode 100644
index 0000000000000000000000000000000000000000..8b1b0075d0866579f1bce80b961a23921c0f0863
--- /dev/null
+++ b/split_info.json
@@ -0,0 +1,28 @@
+{
+ "experiment_name": "exp6_sft_numinamath_dpo",
+ "config_file": "/scratch/jennifer/standard-project-m2-the-transformers/configs/exp6_dpo.yaml",
+ "resolved_config": {
+ "num_train_epochs": 1,
+ "learning_rate": 5e-07,
+ "per_device_train_batch_size": 1,
+ "per_device_eval_batch_size": 1,
+ "gradient_accumulation_steps": 8,
+ "warmup_ratio": 0.1,
+ "seed": 42,
+ "save_total_limit": 3,
+ "eval_steps": 50,
+ "logging_steps": 10,
+ "beta": 0.1,
+ "loss_type": "sigmoid",
+ "max_length": 4096,
+ "max_completion_length": 3072,
+ "val_fraction": 0.1,
+ "wandb_project": "cs552-math-dpo",
+ "experiment_name": "exp6_sft_numinamath_dpo",
+ "output_dir": "/scratch/checkpoints/exp6_sft_numinamath_dpo",
+ "sft_checkpoint": "/scratch/checkpoints/exp6_sft_numinamath",
+ "train_source": "/scratch/data/dpo_pairs_v2/pairs.jsonl"
+ },
+ "n_train": 5048,
+ "n_val": 561
+}
\ No newline at end of file
diff --git a/tokenizer_config.json b/tokenizer_config.json
index 480b16d724b41320e1e217f2408c04a8a8e326fc..c5b200ff35c4189eb88a66e3570d7477243f6739 100644
--- a/tokenizer_config.json
+++ b/tokenizer_config.json
@@ -26,6 +26,5 @@
"pad_token": "<|endoftext|>",
"split_special_tokens": false,
"tokenizer_class": "Qwen2Tokenizer",
- "unk_token": null,
- "chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0].role == 'system' %}\n {{- messages[0].content + '\\n\\n' }}\n {%- endif %}\n {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within XML tags:\\n\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n<|im_end|>\\n\" }}\n{%- else %}\n {#- BOXING INSTRUCTION injected so it is present even when the caller (e.g. CI)\n passes no system message. Targets our no-box / multi-box failure mode. #}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0].role == 'system' %}\n {{- messages[0].content + '\\n\\n' }}\n {%- endif %}\n {{- 'When solving a math problem, work through it step by step and write your final answer inside a \\boxed{...}.<|im_end|>\\n' }}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n {%- set index = (messages|length - 1) - loop.index0 %}\n {%- if ns.multi_step_tool and message.role == \"user\" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %}\n {%- set ns.multi_step_tool = false %}\n {%- set ns.last_query_index = index %}\n {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n {%- if message.content is string %}\n {%- set content = message.content %}\n {%- else %}\n {%- set content = '' %}\n {%- endif %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {%- set reasoning_content = '' %}\n {%- if message.reasoning_content is string %}\n {%- set reasoning_content = message.reasoning_content %}\n {%- else %}\n {%- if '' in content %}\n {%- set reasoning_content = content.split('')[0].rstrip('\\n').split('')[-1].lstrip('\\n') %}\n {%- set content = content.split('')[-1].lstrip('\\n') %}\n {%- endif %}\n {%- endif %}\n {%- if loop.index0 > ns.last_query_index %}\n {%- if loop.last or (not loop.last and reasoning_content) %}\n {{- '<|im_start|>' + message.role + '\\n\\n' + reasoning_content.strip('\\n') + '\\n\\n\\n' + content.lstrip('\\n') }}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- if message.tool_calls %}\n {%- for tool_call in message.tool_calls %}\n {%- if (loop.first and content) or (not loop.first) %}\n {{- '\\n' }}\n {%- endif %}\n {%- if tool_call.function %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {%- if tool_call.arguments is string %}\n {{- tool_call.arguments }}\n {%- else %}\n {{- tool_call.arguments | tojson }}\n {%- endif %}\n {{- '}\\n' }}\n {%- endfor %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n\\n' }}\n {{- content }}\n {{- '\\n' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n {%- if enable_thinking is defined and enable_thinking is false %}\n {{- '\\n\\n\\n\\n' }}\n {%- endif %}\n{%- endif %}"
+ "unk_token": null
}
\ No newline at end of file
diff --git a/trainer_state.json b/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9bd4137a09212f2b1a062e2ff126a39d8970b8cb
--- /dev/null
+++ b/trainer_state.json
@@ -0,0 +1,6389 @@
+{
+ "best_global_step": 750,
+ "best_metric": 1.6413334274291993,
+ "best_model_checkpoint": "/scratch/checkpoints/exp8_grpo_fast/checkpoint-750",
+ "epoch": 0.273224043715847,
+ "eval_steps": 150,
+ "global_step": 750,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333432674407,
+ "completions/max_length": 1388.2,
+ "completions/max_terminated_length": 1149.4,
+ "completions/mean_length": 961.8666870117188,
+ "completions/mean_terminated_length": 813.0800170898438,
+ "completions/min_length": 593.6,
+ "completions/min_terminated_length": 593.6,
+ "entropy": 0.15543196325500805,
+ "epoch": 0.0018214936247723133,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.04320038482546806,
+ "kl": 0.0001638038011151366,
+ "learning_rate": 1.45985401459854e-08,
+ "loss": -0.002662058547139168,
+ "num_tokens": 30920.0,
+ "reward": 2.02333345413208,
+ "reward_std": 0.5894708633422852,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.46666666567325593,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.5233333643525839,
+ "rewards/multi_component_reward/std": 0.4098228693008423,
+ "rewards/no_answer_rate/mean": 0.13333333432674407,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.12532416619360448,
+ "sampling/importance_sampling_ratio/mean": 0.04354772977530956,
+ "sampling/importance_sampling_ratio/min": 0.006865633289925732,
+ "sampling/sampling_logp_difference/max": 0.305552613735199,
+ "sampling/sampling_logp_difference/mean": 0.010744557529687882,
+ "step": 5,
+ "step_time": 32.99731477890164
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1657.0,
+ "completions/max_terminated_length": 1145.8,
+ "completions/mean_length": 919.3000366210938,
+ "completions/mean_terminated_length": 807.0200317382812,
+ "completions/min_length": 442.0,
+ "completions/min_terminated_length": 442.0,
+ "entropy": 0.21022272085150082,
+ "epoch": 0.0036429872495446266,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.014283710159361362,
+ "kl": 0.00032655518104244646,
+ "learning_rate": 3.2846715328467156e-08,
+ "loss": -0.0015700791031122209,
+ "num_tokens": 61163.0,
+ "reward": 1.180000114440918,
+ "reward_std": 0.4240097999572754,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.06666666865348816,
+ "rewards/correctness/std": 0.1632993221282959,
+ "rewards/multi_component_reward/mean": 0.11333334147930145,
+ "rewards/multi_component_reward/std": 0.27013830840587616,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.06362589932978154,
+ "sampling/importance_sampling_ratio/mean": 0.01664090696722269,
+ "sampling/importance_sampling_ratio/min": 0.0005761486502072345,
+ "sampling/sampling_logp_difference/max": 0.49286861419677735,
+ "sampling/sampling_logp_difference/mean": 0.01417195163667202,
+ "step": 10,
+ "step_time": 37.58825172036886
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1657.4,
+ "completions/max_terminated_length": 899.0,
+ "completions/mean_length": 897.2000244140625,
+ "completions/mean_terminated_length": 615.4466735839844,
+ "completions/min_length": 296.4,
+ "completions/min_terminated_length": 296.4,
+ "entropy": 0.20501985649267832,
+ "epoch": 0.00546448087431694,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.0192733071744442,
+ "kl": 0.0003123952769480335,
+ "learning_rate": 5.10948905109489e-08,
+ "loss": -0.004246947914361953,
+ "num_tokens": 93923.0,
+ "reward": 1.8200001120567322,
+ "reward_std": 0.5815625011920929,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.4000000059604645,
+ "rewards/correctness/std": 0.20655910968780516,
+ "rewards/multi_component_reward/mean": 0.42000001221895217,
+ "rewards/multi_component_reward/std": 0.3785122692584991,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.18622441291809083,
+ "sampling/importance_sampling_ratio/mean": 0.04659921806305647,
+ "sampling/importance_sampling_ratio/min": 0.0010122977004923522,
+ "sampling/sampling_logp_difference/max": 0.39253207445144656,
+ "sampling/sampling_logp_difference/mean": 0.01371441949158907,
+ "step": 15,
+ "step_time": 38.31680890209973
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 2023.8,
+ "completions/max_terminated_length": 1348.6,
+ "completions/mean_length": 1195.0000610351562,
+ "completions/mean_terminated_length": 1013.5533630371094,
+ "completions/min_length": 736.4,
+ "completions/min_terminated_length": 736.4,
+ "entropy": 0.15656836417814096,
+ "epoch": 0.007285974499089253,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.02143455669283867,
+ "kl": 0.00023284607256452244,
+ "learning_rate": 6.934306569343065e-08,
+ "loss": -0.006335017085075378,
+ "num_tokens": 131981.0,
+ "reward": 1.3033334732055664,
+ "reward_std": 0.7852452993392944,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.3265986442565918,
+ "rewards/multi_component_reward/mean": 0.1366666793823242,
+ "rewards/multi_component_reward/std": 0.5012870132923126,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.05830998048186302,
+ "sampling/importance_sampling_ratio/mean": 0.016274437960237264,
+ "sampling/importance_sampling_ratio/min": 0.00017535050738298644,
+ "sampling/sampling_logp_difference/max": 0.3216116189956665,
+ "sampling/sampling_logp_difference/mean": 0.011110224667936564,
+ "step": 20,
+ "step_time": 45.75825558342039
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1375.4,
+ "completions/max_terminated_length": 1177.4,
+ "completions/mean_length": 1042.1333740234375,
+ "completions/mean_terminated_length": 911.9000244140625,
+ "completions/min_length": 680.6,
+ "completions/min_terminated_length": 680.6,
+ "entropy": 0.1741524614393711,
+ "epoch": 0.009107468123861567,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.01335141435265541,
+ "kl": 0.0002629145164974034,
+ "learning_rate": 8.759124087591241e-08,
+ "loss": 0.0026412704959511758,
+ "num_tokens": 166221.0,
+ "reward": 1.6033333897590638,
+ "reward_std": 0.5469928443431854,
+ "rewards/boxed_rate/mean": 0.8666666746139526,
+ "rewards/boxed_rate/std": 0.20655910968780516,
+ "rewards/correctness/mean": 0.26666667461395266,
+ "rewards/correctness/std": 0.20655910968780516,
+ "rewards/multi_component_reward/mean": 0.3033333241939545,
+ "rewards/multi_component_reward/std": 0.36893237233161924,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.20655910968780516,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.04747706279158592,
+ "sampling/importance_sampling_ratio/mean": 0.016832177247852086,
+ "sampling/importance_sampling_ratio/min": 0.00119685190729814,
+ "sampling/sampling_logp_difference/max": 0.3412928104400635,
+ "sampling/sampling_logp_difference/mean": 0.011194901075214148,
+ "step": 25,
+ "step_time": 33.345185896754266
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2666666716337204,
+ "completions/max_length": 1851.2,
+ "completions/max_terminated_length": 1397.6,
+ "completions/mean_length": 1248.2000366210937,
+ "completions/mean_terminated_length": 1000.1933471679688,
+ "completions/min_length": 643.4,
+ "completions/min_terminated_length": 643.4,
+ "entropy": 0.1906235640247663,
+ "epoch": 0.01092896174863388,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.02501230686903,
+ "kl": 0.0002681747629443028,
+ "learning_rate": 1.0583941605839415e-07,
+ "loss": -0.0029103375971317293,
+ "num_tokens": 207351.0,
+ "reward": 1.3366667747497558,
+ "reward_std": 0.7248244881629944,
+ "rewards/boxed_rate/mean": 0.7333333313465118,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.1666666716337204,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.10333333685994148,
+ "rewards/multi_component_reward/std": 0.48569379448890687,
+ "rewards/no_answer_rate/mean": 0.2666666716337204,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.05718064345419407,
+ "sampling/importance_sampling_ratio/mean": 0.01394880290608853,
+ "sampling/importance_sampling_ratio/min": 3.71510857371279e-06,
+ "sampling/sampling_logp_difference/max": 0.39807581901550293,
+ "sampling/sampling_logp_difference/mean": 0.011931476183235645,
+ "step": 30,
+ "step_time": 42.76963838078082
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1265.8,
+ "completions/max_terminated_length": 1010.8,
+ "completions/mean_length": 801.2000183105469,
+ "completions/mean_terminated_length": 710.2333435058594,
+ "completions/min_length": 515.6,
+ "completions/min_terminated_length": 515.6,
+ "entropy": 0.18569878712296486,
+ "epoch": 0.012750455373406194,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.013473724946379662,
+ "kl": 0.00025912571412239536,
+ "learning_rate": 1.240875912408759e-07,
+ "loss": -0.007362464070320129,
+ "num_tokens": 234243.0,
+ "reward": 1.5500000715255737,
+ "reward_std": 0.6336740732192994,
+ "rewards/boxed_rate/mean": 0.9333333373069763,
+ "rewards/boxed_rate/std": 0.10327955484390258,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.3166666716337204,
+ "rewards/multi_component_reward/std": 0.35141916275024415,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.10327955484390258,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.10159578723832965,
+ "sampling/importance_sampling_ratio/mean": 0.03656362611800432,
+ "sampling/importance_sampling_ratio/min": 0.002998881617435645,
+ "sampling/sampling_logp_difference/max": 0.3329684495925903,
+ "sampling/sampling_logp_difference/mean": 0.011683504190295934,
+ "step": 35,
+ "step_time": 29.749523213878273
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333334028720856,
+ "completions/max_length": 1788.4,
+ "completions/max_terminated_length": 1253.4,
+ "completions/mean_length": 1151.96669921875,
+ "completions/mean_terminated_length": 852.2566772460938,
+ "completions/min_length": 632.2,
+ "completions/min_terminated_length": 632.2,
+ "entropy": 0.19733075598875682,
+ "epoch": 0.014571948998178506,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.017843080684542656,
+ "kl": 0.00030036380388385927,
+ "learning_rate": 1.4233576642335764e-07,
+ "loss": 0.0052158843725919725,
+ "num_tokens": 271334.0,
+ "reward": 1.3800000905990601,
+ "reward_std": 0.8547543048858642,
+ "rewards/boxed_rate/mean": 0.7666666686534882,
+ "rewards/boxed_rate/std": 0.2128240704536438,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.35449349880218506,
+ "rewards/multi_component_reward/mean": 0.1800000086426735,
+ "rewards/multi_component_reward/std": 0.5035018444061279,
+ "rewards/no_answer_rate/mean": 0.2333333373069763,
+ "rewards/no_answer_rate/std": 0.2128240704536438,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.05587816089391708,
+ "sampling/importance_sampling_ratio/mean": 0.020892591564916074,
+ "sampling/importance_sampling_ratio/min": 2.3939082747448788e-05,
+ "sampling/sampling_logp_difference/max": 0.42613508701324465,
+ "sampling/sampling_logp_difference/mean": 0.012935483455657959,
+ "step": 40,
+ "step_time": 40.51654889807105
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1250.0,
+ "completions/max_terminated_length": 700.2,
+ "completions/mean_length": 596.7666748046875,
+ "completions/mean_terminated_length": 492.7933410644531,
+ "completions/min_length": 341.8,
+ "completions/min_terminated_length": 341.8,
+ "entropy": 0.22809130648771922,
+ "epoch": 0.01639344262295082,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.06362088024616241,
+ "kl": 0.00031622202028908455,
+ "learning_rate": 1.6058394160583942e-07,
+ "loss": -0.0004649309907108545,
+ "num_tokens": 293263.0,
+ "reward": 1.6900001049041748,
+ "reward_std": 0.6551559090614318,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.3000000089406967,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.3900000035762787,
+ "rewards/multi_component_reward/std": 0.37366979122161864,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.06921428106725216,
+ "sampling/importance_sampling_ratio/mean": 0.029317377135157585,
+ "sampling/importance_sampling_ratio/min": 0.00602934339824146,
+ "sampling/sampling_logp_difference/max": 0.35652687549591067,
+ "sampling/sampling_logp_difference/mean": 0.015356982313096523,
+ "step": 45,
+ "step_time": 28.76009795963764
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1886.2,
+ "completions/max_terminated_length": 1599.0,
+ "completions/mean_length": 1163.9333862304688,
+ "completions/mean_terminated_length": 1057.6466918945312,
+ "completions/min_length": 663.8,
+ "completions/min_terminated_length": 663.8,
+ "entropy": 0.17236283471186956,
+ "epoch": 0.018214936247723135,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.01997266709804535,
+ "kl": 0.0002550489671799975,
+ "learning_rate": 1.7883211678832117e-07,
+ "loss": -0.005824955180287361,
+ "num_tokens": 331739.0,
+ "reward": 1.8566666960716247,
+ "reward_std": 0.7818804442882538,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.36666667461395264,
+ "rewards/correctness/std": 0.3161036252975464,
+ "rewards/multi_component_reward/mean": 0.4233333319425583,
+ "rewards/multi_component_reward/std": 0.4823453426361084,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.054672015644609925,
+ "sampling/importance_sampling_ratio/mean": 0.01720548253506422,
+ "sampling/importance_sampling_ratio/min": 0.00035658250299805546,
+ "sampling/sampling_logp_difference/max": 0.3509830951690674,
+ "sampling/sampling_logp_difference/mean": 0.011117835808545352,
+ "step": 50,
+ "step_time": 43.21503445059061
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.3333333432674408,
+ "completions/max_length": 1744.8,
+ "completions/max_terminated_length": 1236.2,
+ "completions/mean_length": 1297.1666931152345,
+ "completions/mean_terminated_length": 1065.0066833496094,
+ "completions/min_length": 936.6,
+ "completions/min_terminated_length": 936.6,
+ "entropy": 0.12224519066512585,
+ "epoch": 0.020036429872495445,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.021082526072859764,
+ "kl": 0.00019305126891898302,
+ "learning_rate": 1.9708029197080292e-07,
+ "loss": 0.0025430308654904366,
+ "num_tokens": 373144.0,
+ "reward": 1.4600000858306885,
+ "reward_std": 0.7388215482234954,
+ "rewards/boxed_rate/mean": 0.6666666597127915,
+ "rewards/boxed_rate/std": 0.35449349880218506,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.12666668593883515,
+ "rewards/multi_component_reward/std": 0.506528252363205,
+ "rewards/no_answer_rate/mean": 0.33333333134651183,
+ "rewards/no_answer_rate/std": 0.35449349880218506,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.18492921590805053,
+ "sampling/importance_sampling_ratio/max": 0.10442680462729186,
+ "sampling/importance_sampling_ratio/mean": 0.04177726461493876,
+ "sampling/importance_sampling_ratio/min": 0.011532440476071316,
+ "sampling/sampling_logp_difference/max": 0.32466731071472166,
+ "sampling/sampling_logp_difference/mean": 0.007843528036028146,
+ "step": 55,
+ "step_time": 40.733187234401704
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333432674407,
+ "completions/max_length": 1700.0,
+ "completions/max_terminated_length": 1413.4,
+ "completions/mean_length": 923.9000366210937,
+ "completions/mean_terminated_length": 779.9666870117187,
+ "completions/min_length": 489.0,
+ "completions/min_terminated_length": 489.0,
+ "entropy": 0.18838655066986879,
+ "epoch": 0.02185792349726776,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.016334548592567444,
+ "kl": 0.0002971158295016115,
+ "learning_rate": 2.1532846715328465e-07,
+ "loss": 0.0031619951128959655,
+ "num_tokens": 403573.0,
+ "reward": 1.393333411216736,
+ "reward_std": 0.45263335704803465,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.1666666716337204,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.19333333075046538,
+ "rewards/multi_component_reward/std": 0.30624626874923705,
+ "rewards/no_answer_rate/mean": 0.13333333432674407,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.14395143389701842,
+ "sampling/importance_sampling_ratio/mean": 0.04990396222565323,
+ "sampling/importance_sampling_ratio/min": 0.00032090346864226403,
+ "sampling/sampling_logp_difference/max": 0.40355303287506106,
+ "sampling/sampling_logp_difference/mean": 0.012135114334523679,
+ "step": 60,
+ "step_time": 38.38553868718445
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1554.4,
+ "completions/max_terminated_length": 1147.6,
+ "completions/mean_length": 972.8000244140625,
+ "completions/mean_terminated_length": 804.5466735839843,
+ "completions/min_length": 498.2,
+ "completions/min_terminated_length": 498.2,
+ "entropy": 0.1604183206955592,
+ "epoch": 0.023679417122040074,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.00623254245147109,
+ "kl": 0.00027490937888311845,
+ "learning_rate": 2.335766423357664e-07,
+ "loss": -0.005589158460497856,
+ "num_tokens": 436225.0,
+ "reward": 1.490000057220459,
+ "reward_std": 0.6202400684356689,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.19000000208616258,
+ "rewards/multi_component_reward/std": 0.4335045665502548,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.1,
+ "rewards/repetition_rate/std": 0.10954451560974121,
+ "sampling/importance_sampling_ratio/max": 0.0990872667171061,
+ "sampling/importance_sampling_ratio/mean": 0.04436926119960845,
+ "sampling/importance_sampling_ratio/min": 0.004196830893285148,
+ "sampling/sampling_logp_difference/max": 0.297141432762146,
+ "sampling/sampling_logp_difference/mean": 0.010508796572685242,
+ "step": 65,
+ "step_time": 36.07466428950429
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1634.0,
+ "completions/max_terminated_length": 994.8,
+ "completions/mean_length": 744.2666870117188,
+ "completions/mean_terminated_length": 601.7000122070312,
+ "completions/min_length": 374.2,
+ "completions/min_terminated_length": 374.2,
+ "entropy": 0.2318819284439087,
+ "epoch": 0.025500910746812388,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.05441533029079437,
+ "kl": 0.0003520799073157832,
+ "learning_rate": 2.518248175182482e-07,
+ "loss": -0.003310435265302658,
+ "num_tokens": 460845.0,
+ "reward": 1.6933334827423097,
+ "reward_std": 0.7755845502018929,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.3,
+ "rewards/correctness/std": 0.34822853803634646,
+ "rewards/multi_component_reward/mean": 0.36000003293156624,
+ "rewards/multi_component_reward/std": 0.5008985161781311,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.056702688336372375,
+ "sampling/importance_sampling_ratio/mean": 0.019475685060024263,
+ "sampling/importance_sampling_ratio/min": 0.001143362923176032,
+ "sampling/sampling_logp_difference/max": 0.3391279220581055,
+ "sampling/sampling_logp_difference/mean": 0.014528140239417554,
+ "step": 70,
+ "step_time": 36.275424292869864
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1515.2,
+ "completions/max_terminated_length": 845.0,
+ "completions/mean_length": 917.0000244140625,
+ "completions/mean_terminated_length": 678.5633483886719,
+ "completions/min_length": 475.4,
+ "completions/min_terminated_length": 475.4,
+ "entropy": 0.16173700019717216,
+ "epoch": 0.0273224043715847,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.018244028091430664,
+ "kl": 0.00024622106284368785,
+ "learning_rate": 2.700729927007299e-07,
+ "loss": -0.014353486895561218,
+ "num_tokens": 491205.0,
+ "reward": 1.633333432674408,
+ "reward_std": 0.8448223888874054,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.3,
+ "rewards/correctness/std": 0.34822853803634646,
+ "rewards/multi_component_reward/mean": 0.3000000298023224,
+ "rewards/multi_component_reward/std": 0.5207933127880097,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.16285581663250923,
+ "sampling/importance_sampling_ratio/mean": 0.05478177797049284,
+ "sampling/importance_sampling_ratio/min": 0.0013068500558716862,
+ "sampling/sampling_logp_difference/max": 0.3721138000488281,
+ "sampling/sampling_logp_difference/mean": 0.011166188679635525,
+ "step": 75,
+ "step_time": 35.12463851571083
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1582.6,
+ "completions/max_terminated_length": 1313.0,
+ "completions/mean_length": 985.9666870117187,
+ "completions/mean_terminated_length": 935.2533569335938,
+ "completions/min_length": 670.8,
+ "completions/min_terminated_length": 670.8,
+ "entropy": 0.17678433768451213,
+ "epoch": 0.029143897996357013,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.009146427735686302,
+ "kl": 0.0002533862963900901,
+ "learning_rate": 2.8832116788321166e-07,
+ "loss": 0.002100883610546589,
+ "num_tokens": 524180.0,
+ "reward": 1.6633334636688233,
+ "reward_std": 0.8858610212802887,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.3823883533477783,
+ "rewards/multi_component_reward/mean": 0.36333334222435953,
+ "rewards/multi_component_reward/std": 0.473619781434536,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.044011454284191134,
+ "sampling/importance_sampling_ratio/mean": 0.013894069381058216,
+ "sampling/importance_sampling_ratio/min": 0.001124254113574473,
+ "sampling/sampling_logp_difference/max": 0.32026147842407227,
+ "sampling/sampling_logp_difference/mean": 0.01106615299358964,
+ "step": 80,
+ "step_time": 36.22448882814497
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1693.6,
+ "completions/max_terminated_length": 1402.8,
+ "completions/mean_length": 1091.4666870117187,
+ "completions/mean_terminated_length": 998.7600219726562,
+ "completions/min_length": 584.2,
+ "completions/min_terminated_length": 584.2,
+ "entropy": 0.1462914695342382,
+ "epoch": 0.030965391621129327,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.018090059980750084,
+ "kl": 0.00022486963668294873,
+ "learning_rate": 3.065693430656934e-07,
+ "loss": -0.00974438264966011,
+ "num_tokens": 560254.0,
+ "reward": 1.8100001335144043,
+ "reward_std": 1.0823543906211852,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.36666667461395264,
+ "rewards/correctness/std": 0.4794029474258423,
+ "rewards/multi_component_reward/mean": 0.44333334267139435,
+ "rewards/multi_component_reward/std": 0.6124810755252839,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.1276155561208725,
+ "sampling/importance_sampling_ratio/mean": 0.03584536015987396,
+ "sampling/importance_sampling_ratio/min": 0.002319600686418858,
+ "sampling/sampling_logp_difference/max": 0.41160542964935304,
+ "sampling/sampling_logp_difference/mean": 0.010097185987979174,
+ "step": 85,
+ "step_time": 39.475563449971375
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.3333333432674408,
+ "completions/max_length": 1841.2,
+ "completions/max_terminated_length": 875.0,
+ "completions/mean_length": 1059.2000122070312,
+ "completions/mean_terminated_length": 630.0533386230469,
+ "completions/min_length": 458.8,
+ "completions/min_terminated_length": 458.8,
+ "entropy": 0.2378736046453317,
+ "epoch": 0.03278688524590164,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.00910874642431736,
+ "kl": 0.0003503820747331095,
+ "learning_rate": 3.2481751824817516e-07,
+ "loss": 0.0034383241087198257,
+ "num_tokens": 593668.0,
+ "reward": 1.4166667222976685,
+ "reward_std": 0.8338384032249451,
+ "rewards/boxed_rate/mean": 0.6333333253860474,
+ "rewards/boxed_rate/std": 0.451508092880249,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.11666668206453323,
+ "rewards/multi_component_reward/std": 0.566602936387062,
+ "rewards/no_answer_rate/mean": 0.36666667759418486,
+ "rewards/no_answer_rate/std": 0.451508092880249,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.05201358497142792,
+ "sampling/importance_sampling_ratio/mean": 0.017090958543121815,
+ "sampling/importance_sampling_ratio/min": 0.0002041866974779629,
+ "sampling/sampling_logp_difference/max": 0.3399634838104248,
+ "sampling/sampling_logp_difference/mean": 0.015104132890701293,
+ "step": 90,
+ "step_time": 41.00809515919536
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1441.6,
+ "completions/max_terminated_length": 960.8,
+ "completions/mean_length": 802.6000366210938,
+ "completions/mean_terminated_length": 711.3800231933594,
+ "completions/min_length": 525.6,
+ "completions/min_terminated_length": 525.6,
+ "entropy": 0.2225553606947263,
+ "epoch": 0.03460837887067395,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.01798272132873535,
+ "kl": 0.00032260040970868433,
+ "learning_rate": 3.4306569343065697e-07,
+ "loss": 0.007113450020551681,
+ "num_tokens": 621316.0,
+ "reward": 1.223333477973938,
+ "reward_std": 0.36713925525546076,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.06666666865348816,
+ "rewards/correctness/std": 0.1632993221282959,
+ "rewards/multi_component_reward/mean": 0.12333333715796471,
+ "rewards/multi_component_reward/std": 0.2738735854625702,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.15052475407719612,
+ "sampling/importance_sampling_ratio/mean": 0.046577111538499597,
+ "sampling/importance_sampling_ratio/min": 0.012816431716431823,
+ "sampling/sampling_logp_difference/max": 0.34408046007156373,
+ "sampling/sampling_logp_difference/mean": 0.014901423826813698,
+ "step": 95,
+ "step_time": 32.91087112892419
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1537.8,
+ "completions/max_terminated_length": 1380.2,
+ "completions/mean_length": 972.9000366210937,
+ "completions/mean_terminated_length": 902.8400146484375,
+ "completions/min_length": 618.0,
+ "completions/min_terminated_length": 618.0,
+ "entropy": 0.2104227361579736,
+ "epoch": 0.03642987249544627,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.006778502371162176,
+ "kl": 0.00031352789907638606,
+ "learning_rate": 3.6131386861313867e-07,
+ "loss": -0.0013602237217128278,
+ "num_tokens": 653443.0,
+ "reward": 1.880000078678131,
+ "reward_std": 0.9815195143222809,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.4000000059604645,
+ "rewards/correctness/std": 0.4256481409072876,
+ "rewards/multi_component_reward/mean": 0.48000000715255736,
+ "rewards/multi_component_reward/std": 0.5586783826351166,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.0448111507575959,
+ "sampling/importance_sampling_ratio/mean": 0.01701541549991816,
+ "sampling/importance_sampling_ratio/min": 0.0003750977984182488,
+ "sampling/sampling_logp_difference/max": 0.3470792770385742,
+ "sampling/sampling_logp_difference/mean": 0.013565080799162388,
+ "step": 100,
+ "step_time": 35.31437961217016
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1421.4,
+ "completions/max_terminated_length": 1072.0,
+ "completions/mean_length": 947.7000366210938,
+ "completions/mean_terminated_length": 787.706689453125,
+ "completions/min_length": 551.6,
+ "completions/min_terminated_length": 551.6,
+ "entropy": 0.2239283885806799,
+ "epoch": 0.03825136612021858,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.011714774183928967,
+ "kl": 0.0003353157648234628,
+ "learning_rate": 3.795620437956204e-07,
+ "loss": 0.009437119215726852,
+ "num_tokens": 683674.0,
+ "reward": 1.4433334589004516,
+ "reward_std": 0.8665581226348877,
+ "rewards/boxed_rate/mean": 0.8333333313465119,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.21000000834465027,
+ "rewards/multi_component_reward/std": 0.5039317846298218,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.1078285550349392,
+ "sampling/importance_sampling_ratio/mean": 0.03690805228543468,
+ "sampling/importance_sampling_ratio/min": 0.0026856323238462275,
+ "sampling/sampling_logp_difference/max": 0.41998746395111086,
+ "sampling/sampling_logp_difference/mean": 0.014628523960709572,
+ "step": 105,
+ "step_time": 32.93989271316677
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1471.4,
+ "completions/max_terminated_length": 979.4,
+ "completions/mean_length": 852.2666748046875,
+ "completions/mean_terminated_length": 721.6166748046875,
+ "completions/min_length": 488.4,
+ "completions/min_terminated_length": 488.4,
+ "entropy": 0.14421232057114441,
+ "epoch": 0.04007285974499089,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.07266412675380707,
+ "kl": 0.00022867745647090488,
+ "learning_rate": 3.978102189781022e-07,
+ "loss": 0.0018386229872703551,
+ "num_tokens": 713286.0,
+ "reward": 1.6733334064483643,
+ "reward_std": 0.7425484180450439,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.3000000059604645,
+ "rewards/correctness/std": 0.3161036252975464,
+ "rewards/multi_component_reward/mean": 0.3400000125169754,
+ "rewards/multi_component_reward/std": 0.47312636375427247,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.13587955459952356,
+ "sampling/importance_sampling_ratio/mean": 0.050121305510401726,
+ "sampling/importance_sampling_ratio/min": 0.0025831204319047172,
+ "sampling/sampling_logp_difference/max": 0.345587694644928,
+ "sampling/sampling_logp_difference/mean": 0.00959718506783247,
+ "step": 110,
+ "step_time": 33.89908826816827
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2333333373069763,
+ "completions/max_length": 1749.4,
+ "completions/max_terminated_length": 884.2,
+ "completions/mean_length": 869.533349609375,
+ "completions/mean_terminated_length": 539.5833374023438,
+ "completions/min_length": 307.8,
+ "completions/min_terminated_length": 307.8,
+ "entropy": 0.2536199669043223,
+ "epoch": 0.04189435336976321,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.011132585816085339,
+ "kl": 0.0003656693840942656,
+ "learning_rate": 4.160583941605839e-07,
+ "loss": 0.0021231153979897497,
+ "num_tokens": 741952.0,
+ "reward": 1.7500001192092896,
+ "reward_std": 1.106396782398224,
+ "rewards/boxed_rate/mean": 0.799999988079071,
+ "rewards/boxed_rate/std": 0.35449349880218506,
+ "rewards/correctness/mean": 0.36666667461395264,
+ "rewards/correctness/std": 0.4794029474258423,
+ "rewards/multi_component_reward/mean": 0.383333346247673,
+ "rewards/multi_component_reward/std": 0.6433047533035279,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.35449349880218506,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.25683521628379824,
+ "sampling/importance_sampling_ratio/mean": 0.065315605327487,
+ "sampling/importance_sampling_ratio/min": 0.00023330498968689221,
+ "sampling/sampling_logp_difference/max": 0.35823286771774293,
+ "sampling/sampling_logp_difference/mean": 0.01769500244408846,
+ "step": 115,
+ "step_time": 39.532602636888626
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1573.0,
+ "completions/max_terminated_length": 1324.8,
+ "completions/mean_length": 1093.2000366210937,
+ "completions/mean_terminated_length": 1037.0133666992188,
+ "completions/min_length": 766.8,
+ "completions/min_terminated_length": 766.8,
+ "entropy": 0.15541148359576862,
+ "epoch": 0.04371584699453552,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.01879267208278179,
+ "kl": 0.00025834220529456314,
+ "learning_rate": 4.343065693430657e-07,
+ "loss": 0.0006856221705675125,
+ "num_tokens": 777766.0,
+ "reward": 1.433333396911621,
+ "reward_std": 0.4934782743453979,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2128240704536438,
+ "rewards/multi_component_reward/mean": 0.23333334028720856,
+ "rewards/multi_component_reward/std": 0.32306827008724215,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.027404608484357596,
+ "sampling/importance_sampling_ratio/mean": 0.012392069399356841,
+ "sampling/importance_sampling_ratio/min": 0.0004944111146869545,
+ "sampling/sampling_logp_difference/max": 0.37873687744140627,
+ "sampling/sampling_logp_difference/mean": 0.010291843488812447,
+ "step": 120,
+ "step_time": 36.481429217197004
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1672.2,
+ "completions/max_terminated_length": 1231.2,
+ "completions/mean_length": 1009.0000122070312,
+ "completions/mean_terminated_length": 875.0133422851562,
+ "completions/min_length": 573.0,
+ "completions/min_terminated_length": 573.0,
+ "entropy": 0.16983376170198122,
+ "epoch": 0.04553734061930783,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.026680519804358482,
+ "kl": 0.00028378382169951997,
+ "learning_rate": 4.5255474452554743e-07,
+ "loss": 0.002333539165556431,
+ "num_tokens": 811096.0,
+ "reward": 1.3900000691413879,
+ "reward_std": 0.4993088662624359,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.1666666716337204,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.22333332896232605,
+ "rewards/multi_component_reward/std": 0.31437968015670775,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.1588123269379139,
+ "sampling/importance_sampling_ratio/mean": 0.04118972420692444,
+ "sampling/importance_sampling_ratio/min": 0.002068098068754896,
+ "sampling/sampling_logp_difference/max": 0.42397408485412597,
+ "sampling/sampling_logp_difference/mean": 0.01178776090964675,
+ "step": 125,
+ "step_time": 38.221346308663485
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1774.8,
+ "completions/max_terminated_length": 1017.6,
+ "completions/mean_length": 970.7666870117188,
+ "completions/mean_terminated_length": 722.2733520507812,
+ "completions/min_length": 472.0,
+ "completions/min_terminated_length": 472.0,
+ "entropy": 0.20842353130380312,
+ "epoch": 0.04735883424408015,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.017246264964342117,
+ "kl": 0.00032587918249191714,
+ "learning_rate": 4.7080291970802913e-07,
+ "loss": -0.004489587619900703,
+ "num_tokens": 843681.0,
+ "reward": 1.3233334302902222,
+ "reward_std": 0.6886623933911323,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.12333333268761634,
+ "rewards/multi_component_reward/std": 0.4415378957986832,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.057003194093704225,
+ "sampling/importance_sampling_ratio/mean": 0.015649724379181863,
+ "sampling/importance_sampling_ratio/min": 0.0008366983823620787,
+ "sampling/sampling_logp_difference/max": 0.42825262546539306,
+ "sampling/sampling_logp_difference/mean": 0.013275405205786229,
+ "step": 130,
+ "step_time": 40.03070425353944
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1638.0,
+ "completions/max_terminated_length": 1082.6,
+ "completions/mean_length": 965.3000427246094,
+ "completions/mean_terminated_length": 805.8833435058593,
+ "completions/min_length": 475.4,
+ "completions/min_terminated_length": 475.4,
+ "entropy": 0.20910785247882208,
+ "epoch": 0.04918032786885246,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.014268649742007256,
+ "kl": 0.00030581061485766744,
+ "learning_rate": 4.89051094890511e-07,
+ "loss": -0.0031251441687345505,
+ "num_tokens": 875316.0,
+ "reward": 1.3700001120567322,
+ "reward_std": 0.6670311987400055,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.20333334803581238,
+ "rewards/multi_component_reward/std": 0.4036152184009552,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.051390893710777166,
+ "sampling/importance_sampling_ratio/mean": 0.015738325094571338,
+ "sampling/importance_sampling_ratio/min": 0.0017150626291213885,
+ "sampling/sampling_logp_difference/max": 0.3242307841777802,
+ "sampling/sampling_logp_difference/mean": 0.01380961835384369,
+ "step": 135,
+ "step_time": 37.537090591713785
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333334028720856,
+ "completions/max_length": 1763.0,
+ "completions/max_terminated_length": 775.6,
+ "completions/mean_length": 960.7000122070312,
+ "completions/mean_terminated_length": 604.2866760253906,
+ "completions/min_length": 455.4,
+ "completions/min_terminated_length": 455.4,
+ "entropy": 0.23928763965765634,
+ "epoch": 0.051001821493624776,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.037585072219371796,
+ "kl": 0.0004156571065929408,
+ "learning_rate": 4.996165644171779e-07,
+ "loss": -0.003277498111128807,
+ "num_tokens": 907599.0,
+ "reward": 1.8700001239776611,
+ "reward_std": 1.213301706314087,
+ "rewards/boxed_rate/mean": 0.7666666567325592,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.4333333343267441,
+ "rewards/correctness/std": 0.5198277235031128,
+ "rewards/multi_component_reward/mean": 0.4366666838526726,
+ "rewards/multi_component_reward/std": 0.7051358222961426,
+ "rewards/no_answer_rate/mean": 0.23333334028720856,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.03320965538732708,
+ "sampling/importance_sampling_ratio/mean": 0.015124964842107147,
+ "sampling/importance_sampling_ratio/min": 0.0001439221122021143,
+ "sampling/sampling_logp_difference/max": 0.3350650787353516,
+ "sampling/sampling_logp_difference/mean": 0.015135842747986317,
+ "step": 140,
+ "step_time": 39.69141132887453
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333432674407,
+ "completions/max_length": 1378.6,
+ "completions/max_terminated_length": 1035.8,
+ "completions/mean_length": 898.6333557128906,
+ "completions/mean_terminated_length": 752.8266784667969,
+ "completions/min_length": 510.2,
+ "completions/min_terminated_length": 510.2,
+ "entropy": 0.15787406551341215,
+ "epoch": 0.052823315118397086,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.19890737533569336,
+ "kl": 0.0003336386442242656,
+ "learning_rate": 4.986579754601227e-07,
+ "loss": 0.002312604896724224,
+ "num_tokens": 936976.0,
+ "reward": 1.6733333706855773,
+ "reward_std": 0.7778201699256897,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.3000000059604645,
+ "rewards/correctness/std": 0.3161036252975464,
+ "rewards/multi_component_reward/mean": 0.34000000953674314,
+ "rewards/multi_component_reward/std": 0.4764533966779709,
+ "rewards/no_answer_rate/mean": 0.13333333432674407,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.17277967557311058,
+ "sampling/importance_sampling_ratio/mean": 0.0497883933596313,
+ "sampling/importance_sampling_ratio/min": 0.0009090198537052175,
+ "sampling/sampling_logp_difference/max": 0.5492818593978882,
+ "sampling/sampling_logp_difference/mean": 0.010555424820631742,
+ "step": 145,
+ "step_time": 32.12684296686202
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1774.2,
+ "completions/max_terminated_length": 1617.8,
+ "completions/mean_length": 1212.6333618164062,
+ "completions/mean_terminated_length": 1111.5633544921875,
+ "completions/min_length": 845.2,
+ "completions/min_terminated_length": 845.2,
+ "entropy": 0.17395392432808876,
+ "epoch": 0.0546448087431694,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.021323509514331818,
+ "kl": 0.00030080197708836444,
+ "learning_rate": 4.976993865030675e-07,
+ "loss": -0.005566118285059929,
+ "num_tokens": 976577.0,
+ "reward": 1.4133334517478944,
+ "reward_std": 0.5511920034885407,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2128240704536438,
+ "rewards/multi_component_reward/mean": 0.17999999672174455,
+ "rewards/multi_component_reward/std": 0.3554683744907379,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.05498660355806351,
+ "sampling/importance_sampling_ratio/mean": 0.01640337195713073,
+ "sampling/importance_sampling_ratio/min": 1.4973960100639972e-05,
+ "sampling/sampling_logp_difference/max": 0.47727389335632325,
+ "sampling/sampling_logp_difference/mean": 0.011420094780623913,
+ "step": 150,
+ "step_time": 41.61420878618956
+ },
+ {
+ "epoch": 0.0546448087431694,
+ "eval_clip_ratio/high_max": 0.0,
+ "eval_clip_ratio/high_mean": 0.0,
+ "eval_clip_ratio/low_mean": 0.0,
+ "eval_clip_ratio/low_min": 0.0,
+ "eval_clip_ratio/region_mean": 0.0,
+ "eval_completions/clipped_ratio": 0.21666667133569717,
+ "eval_completions/max_length": 1789.4,
+ "eval_completions/max_terminated_length": 1216.88,
+ "eval_completions/mean_length": 1121.4300323486327,
+ "eval_completions/mean_terminated_length": 886.213013305664,
+ "eval_completions/min_length": 641.78,
+ "eval_completions/min_terminated_length": 641.78,
+ "eval_entropy": 0.17698929071426392,
+ "eval_frac_reward_zero_std": 0.08,
+ "eval_kl": 0.00028998344379942864,
+ "eval_loss": -0.0026245773769915104,
+ "eval_num_tokens": 976577.0,
+ "eval_reward": 1.5393334233760834,
+ "eval_reward_std": 0.7298788775503635,
+ "eval_rewards/boxed_rate/mean": 0.7833333307504654,
+ "eval_rewards/boxed_rate/std": 0.32846659898757935,
+ "eval_rewards/correctness/mean": 0.24666666924953462,
+ "eval_rewards/correctness/std": 0.27019834876060483,
+ "eval_rewards/multi_component_reward/mean": 0.21933334972709417,
+ "eval_rewards/multi_component_reward/std": 0.48081229120492935,
+ "eval_rewards/no_answer_rate/mean": 0.21666667133569717,
+ "eval_rewards/no_answer_rate/std": 0.32846659898757935,
+ "eval_rewards/repetition_rate/mean": 0.07333333462476731,
+ "eval_rewards/repetition_rate/std": 0.13300593733787536,
+ "eval_runtime": 1782.1981,
+ "eval_samples_per_second": 0.028,
+ "eval_sampling/importance_sampling_ratio/max": 0.07466922109248117,
+ "eval_sampling/importance_sampling_ratio/mean": 0.026714042400126346,
+ "eval_sampling/importance_sampling_ratio/min": 0.0011027993811156295,
+ "eval_sampling/sampling_logp_difference/max": 0.4103265881538391,
+ "eval_sampling/sampling_logp_difference/mean": 0.011447538509964944,
+ "eval_steps_per_second": 0.005,
+ "step": 150
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.0,
+ "completions/max_length": 1207.0,
+ "completions/max_terminated_length": 1207.0,
+ "completions/mean_length": 770.1333679199219,
+ "completions/mean_terminated_length": 770.1333679199219,
+ "completions/min_length": 500.8,
+ "completions/min_terminated_length": 500.8,
+ "entropy": 0.19859646161397299,
+ "epoch": 0.056466302367941715,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.05886993184685707,
+ "kl": 0.0003771214178414084,
+ "learning_rate": 4.967407975460122e-07,
+ "loss": -0.004334203898906708,
+ "num_tokens": 1003269.0,
+ "reward": 1.4733334302902221,
+ "reward_std": 0.504085260629654,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2128240704536438,
+ "rewards/multi_component_reward/mean": 0.27333334758877753,
+ "rewards/multi_component_reward/std": 0.2586013779044151,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.10831067697145044,
+ "sampling/importance_sampling_ratio/mean": 0.040180114656686784,
+ "sampling/importance_sampling_ratio/min": 0.0008547768986318261,
+ "sampling/sampling_logp_difference/max": 0.3474395990371704,
+ "sampling/sampling_logp_difference/mean": 0.013024460710585117,
+ "step": 155,
+ "step_time": 28.411310048028827
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1885.8,
+ "completions/max_terminated_length": 1037.6,
+ "completions/mean_length": 888.3666748046875,
+ "completions/mean_terminated_length": 646.7700012207031,
+ "completions/min_length": 398.6,
+ "completions/min_terminated_length": 398.6,
+ "entropy": 0.21153039361039797,
+ "epoch": 0.058287795992714025,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.030352359637618065,
+ "kl": 0.00036930939143834014,
+ "learning_rate": 4.957822085889571e-07,
+ "loss": -0.015004897117614746,
+ "num_tokens": 1032692.0,
+ "reward": 1.6533334136009217,
+ "reward_std": 0.7879160098731518,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.3000000059604645,
+ "rewards/correctness/std": 0.3161036252975464,
+ "rewards/multi_component_reward/mean": 0.3200000114738941,
+ "rewards/multi_component_reward/std": 0.5426497280597686,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.15256396904587746,
+ "sampling/importance_sampling_ratio/mean": 0.05275641949847341,
+ "sampling/importance_sampling_ratio/min": 0.0003334702798868202,
+ "sampling/sampling_logp_difference/max": 0.3136852741241455,
+ "sampling/sampling_logp_difference/mean": 0.014769792556762695,
+ "step": 160,
+ "step_time": 41.816410617157814
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2666666716337204,
+ "completions/max_length": 1347.0,
+ "completions/max_terminated_length": 859.2,
+ "completions/mean_length": 907.3000244140625,
+ "completions/mean_terminated_length": 642.22001953125,
+ "completions/min_length": 436.6,
+ "completions/min_terminated_length": 436.6,
+ "entropy": 0.2138912024597327,
+ "epoch": 0.060109289617486336,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.0335945226252079,
+ "kl": 0.00036641353217419236,
+ "learning_rate": 4.948236196319019e-07,
+ "loss": -0.001109707448631525,
+ "num_tokens": 1062455.0,
+ "reward": 1.733333444595337,
+ "reward_std": 0.6965994656085968,
+ "rewards/boxed_rate/mean": 0.7333333313465118,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.3666666686534882,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.33333335630595684,
+ "rewards/multi_component_reward/std": 0.4374739408493042,
+ "rewards/no_answer_rate/mean": 0.2666666716337204,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.17867551841773094,
+ "sampling/importance_sampling_ratio/mean": 0.07929584998637437,
+ "sampling/importance_sampling_ratio/min": 0.007277285696347713,
+ "sampling/sampling_logp_difference/max": 0.3139556646347046,
+ "sampling/sampling_logp_difference/mean": 0.014788956940174102,
+ "step": 165,
+ "step_time": 31.849167810566723
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1693.4,
+ "completions/max_terminated_length": 1086.8,
+ "completions/mean_length": 985.8000244140625,
+ "completions/mean_terminated_length": 765.3133666992187,
+ "completions/min_length": 539.0,
+ "completions/min_terminated_length": 539.0,
+ "entropy": 0.22184887006878853,
+ "epoch": 0.061930783242258654,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.010766901075839996,
+ "kl": 0.0003837254077855808,
+ "learning_rate": 4.938650306748465e-07,
+ "loss": -0.005718140304088593,
+ "num_tokens": 1094465.0,
+ "reward": 1.5600001335144043,
+ "reward_std": 0.6616819977760315,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.2666666656732559,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.2933333650231361,
+ "rewards/multi_component_reward/std": 0.4033259034156799,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.04150749985128641,
+ "sampling/importance_sampling_ratio/mean": 0.017618318973109125,
+ "sampling/importance_sampling_ratio/min": 0.0001386483045517349,
+ "sampling/sampling_logp_difference/max": 0.41498665809631347,
+ "sampling/sampling_logp_difference/mean": 0.015049760602414608,
+ "step": 170,
+ "step_time": 38.327568624168634
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.33333334028720857,
+ "completions/max_length": 2048.0,
+ "completions/max_terminated_length": 1280.8,
+ "completions/mean_length": 1264.76669921875,
+ "completions/mean_terminated_length": 950.460009765625,
+ "completions/min_length": 712.6,
+ "completions/min_terminated_length": 712.6,
+ "entropy": 0.20655053692559402,
+ "epoch": 0.06375227686703097,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.1300213485956192,
+ "kl": 0.00040407002428158496,
+ "learning_rate": 4.929064417177914e-07,
+ "loss": -0.002616160549223423,
+ "num_tokens": 1134790.0,
+ "reward": 1.2733334183692933,
+ "reward_std": 0.7226614370942116,
+ "rewards/boxed_rate/mean": 0.6666666567325592,
+ "rewards/boxed_rate/std": 0.45777305364608767,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.0733333457261324,
+ "rewards/multi_component_reward/std": 0.5279393553733825,
+ "rewards/no_answer_rate/mean": 0.33333334028720857,
+ "rewards/no_answer_rate/std": 0.45777305364608767,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.05303769689053297,
+ "sampling/importance_sampling_ratio/mean": 0.015054715669248254,
+ "sampling/importance_sampling_ratio/min": 8.732383148210488e-05,
+ "sampling/sampling_logp_difference/max": 0.29345667362213135,
+ "sampling/sampling_logp_difference/mean": 0.013776615634560585,
+ "step": 175,
+ "step_time": 46.28360453415662
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2333333373069763,
+ "completions/max_length": 1848.8,
+ "completions/max_terminated_length": 1419.8,
+ "completions/mean_length": 1218.3667114257812,
+ "completions/mean_terminated_length": 983.3333618164063,
+ "completions/min_length": 721.0,
+ "completions/min_terminated_length": 721.0,
+ "entropy": 0.14731517334779104,
+ "epoch": 0.06557377049180328,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.03868393227458,
+ "kl": 0.00034544099568544574,
+ "learning_rate": 4.919478527607362e-07,
+ "loss": -0.007346557080745697,
+ "num_tokens": 1175133.0,
+ "reward": 1.7500000834465026,
+ "reward_std": 0.9921784222126007,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.3666666716337204,
+ "rewards/correctness/std": 0.404018247127533,
+ "rewards/multi_component_reward/mean": 0.383333346247673,
+ "rewards/multi_component_reward/std": 0.5928741514682769,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.08876338973641396,
+ "sampling/importance_sampling_ratio/mean": 0.02539586406201124,
+ "sampling/importance_sampling_ratio/min": 0.0011980376816850713,
+ "sampling/sampling_logp_difference/max": 0.3423057317733765,
+ "sampling/sampling_logp_difference/mean": 0.010648915357887746,
+ "step": 180,
+ "step_time": 42.86893743276596
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000596046447,
+ "completions/max_length": 1706.6,
+ "completions/max_terminated_length": 1151.0,
+ "completions/mean_length": 1097.7666870117187,
+ "completions/mean_terminated_length": 885.9666931152344,
+ "completions/min_length": 678.4,
+ "completions/min_terminated_length": 678.4,
+ "entropy": 0.13618133092919985,
+ "epoch": 0.06739526411657559,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.04725215584039688,
+ "kl": 0.0003020782351086382,
+ "learning_rate": 4.909892638036809e-07,
+ "loss": -0.0009320625104010105,
+ "num_tokens": 1210850.0,
+ "reward": 1.5866667985916139,
+ "reward_std": 0.3331963121891022,
+ "rewards/boxed_rate/mean": 0.7999999940395355,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.10327955484390258,
+ "rewards/multi_component_reward/mean": 0.2533333495259285,
+ "rewards/multi_component_reward/std": 0.27605399787425994,
+ "rewards/no_answer_rate/mean": 0.20000000596046447,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.0795806871727109,
+ "sampling/importance_sampling_ratio/mean": 0.02610416170209646,
+ "sampling/importance_sampling_ratio/min": 0.002741426149717241,
+ "sampling/sampling_logp_difference/max": 0.3459432005882263,
+ "sampling/sampling_logp_difference/mean": 0.008953130897134542,
+ "step": 185,
+ "step_time": 39.274089214019476
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1897.2,
+ "completions/max_terminated_length": 1270.2,
+ "completions/mean_length": 1048.533349609375,
+ "completions/mean_terminated_length": 838.5800048828125,
+ "completions/min_length": 490.8,
+ "completions/min_terminated_length": 490.8,
+ "entropy": 0.18140921418865522,
+ "epoch": 0.0692167577413479,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.05130758509039879,
+ "kl": 0.00038463542005047203,
+ "learning_rate": 4.900306748466257e-07,
+ "loss": -0.00950215756893158,
+ "num_tokens": 1245174.0,
+ "reward": 1.2133333802223205,
+ "reward_std": 0.6049932837486267,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.36985843181610106,
+ "rewards/correctness/mean": 0.10000000298023223,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.08000001087784767,
+ "rewards/multi_component_reward/std": 0.39686688780784607,
+ "rewards/no_answer_rate/mean": 0.20000000596046447,
+ "rewards/no_answer_rate/std": 0.36985843181610106,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.0888688538223505,
+ "sampling/importance_sampling_ratio/mean": 0.03125911168754101,
+ "sampling/importance_sampling_ratio/min": 0.00010974614788835727,
+ "sampling/sampling_logp_difference/max": 0.3610305070877075,
+ "sampling/sampling_logp_difference/mean": 0.01223631165921688,
+ "step": 190,
+ "step_time": 42.93450187277049
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2333333373069763,
+ "completions/max_length": 1931.4,
+ "completions/max_terminated_length": 1245.0,
+ "completions/mean_length": 1162.5333740234375,
+ "completions/mean_terminated_length": 899.483349609375,
+ "completions/min_length": 618.2,
+ "completions/min_terminated_length": 618.2,
+ "entropy": 0.1432020000492533,
+ "epoch": 0.07103825136612021,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.06832960247993469,
+ "kl": 0.00033801408014066204,
+ "learning_rate": 4.890720858895706e-07,
+ "loss": -0.00368819460272789,
+ "num_tokens": 1282900.0,
+ "reward": 1.6600000619888307,
+ "reward_std": 1.03403559923172,
+ "rewards/boxed_rate/mean": 0.7666666626930236,
+ "rewards/boxed_rate/std": 0.3761233925819397,
+ "rewards/correctness/mean": 0.33333333432674406,
+ "rewards/correctness/std": 0.4298781991004944,
+ "rewards/multi_component_reward/mean": 0.3266666799783707,
+ "rewards/multi_component_reward/std": 0.6201063513755798,
+ "rewards/no_answer_rate/mean": 0.2333333373069763,
+ "rewards/no_answer_rate/std": 0.3761233925819397,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.24176665293052793,
+ "sampling/importance_sampling_ratio/mean": 0.05982898166403174,
+ "sampling/importance_sampling_ratio/min": 0.010627524812724437,
+ "sampling/sampling_logp_difference/max": 0.3079464197158813,
+ "sampling/sampling_logp_difference/mean": 0.009482560632750391,
+ "step": 195,
+ "step_time": 43.80696959905326
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1874.4,
+ "completions/max_terminated_length": 1207.6,
+ "completions/mean_length": 1102.2666748046875,
+ "completions/mean_terminated_length": 956.0200134277344,
+ "completions/min_length": 668.8,
+ "completions/min_terminated_length": 668.8,
+ "entropy": 0.13664823652555544,
+ "epoch": 0.07285974499089254,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.01865367591381073,
+ "kl": 0.00029515944139954324,
+ "learning_rate": 4.881134969325153e-07,
+ "loss": 0.006076012551784515,
+ "num_tokens": 1318920.0,
+ "reward": 1.9533334493637085,
+ "reward_std": 1.1985347509384154,
+ "rewards/boxed_rate/mean": 0.8666666507720947,
+ "rewards/boxed_rate/std": 0.3265986442565918,
+ "rewards/correctness/mean": 0.4333333373069763,
+ "rewards/correctness/std": 0.5351926565170289,
+ "rewards/multi_component_reward/mean": 0.4866666853427887,
+ "rewards/multi_component_reward/std": 0.7141769766807556,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.3265986442565918,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.08207609243690968,
+ "sampling/importance_sampling_ratio/mean": 0.03297081319615245,
+ "sampling/importance_sampling_ratio/min": 0.003100475773135258,
+ "sampling/sampling_logp_difference/max": 0.40416555404663085,
+ "sampling/sampling_logp_difference/mean": 0.008233289048075677,
+ "step": 200,
+ "step_time": 42.36675942577422
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1600.0,
+ "completions/max_terminated_length": 1066.6,
+ "completions/mean_length": 959.2000122070312,
+ "completions/mean_terminated_length": 798.0866943359375,
+ "completions/min_length": 461.0,
+ "completions/min_terminated_length": 461.0,
+ "entropy": 0.14408907170097032,
+ "epoch": 0.07468123861566485,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.032731793820858,
+ "kl": 0.0003691373499653613,
+ "learning_rate": 4.871549079754601e-07,
+ "loss": -0.001960751600563526,
+ "num_tokens": 1352016.0,
+ "reward": 1.673333430290222,
+ "reward_std": 0.9306211471557617,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.3000000029802322,
+ "rewards/correctness/std": 0.404018247127533,
+ "rewards/multi_component_reward/mean": 0.3400000214576721,
+ "rewards/multi_component_reward/std": 0.5803858906030654,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.10455413907766342,
+ "sampling/importance_sampling_ratio/mean": 0.03612115085124969,
+ "sampling/importance_sampling_ratio/min": 0.0026871272621811215,
+ "sampling/sampling_logp_difference/max": 0.3213069200515747,
+ "sampling/sampling_logp_difference/mean": 0.00922326734289527,
+ "step": 205,
+ "step_time": 37.272668573819104
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1572.2,
+ "completions/max_terminated_length": 1133.2,
+ "completions/mean_length": 879.3333618164063,
+ "completions/mean_terminated_length": 743.8000244140625,
+ "completions/min_length": 459.6,
+ "completions/min_terminated_length": 459.6,
+ "entropy": 0.19108737086256344,
+ "epoch": 0.07650273224043716,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.018986158072948456,
+ "kl": 0.0004163069315836765,
+ "learning_rate": 4.86196319018405e-07,
+ "loss": 0.0035621844232082366,
+ "num_tokens": 1380880.0,
+ "reward": 1.740000104904175,
+ "reward_std": 0.947225558757782,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.33333333432674406,
+ "rewards/correctness/std": 0.4298781991004944,
+ "rewards/multi_component_reward/mean": 0.40666669309139253,
+ "rewards/multi_component_reward/std": 0.5280151665210724,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.07124827802181244,
+ "sampling/importance_sampling_ratio/mean": 0.02541533587500453,
+ "sampling/importance_sampling_ratio/min": 0.002891992630936181,
+ "sampling/sampling_logp_difference/max": 0.3407265663146973,
+ "sampling/sampling_logp_difference/mean": 0.012478712201118469,
+ "step": 210,
+ "step_time": 35.72363304812461
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1773.4,
+ "completions/max_terminated_length": 1327.6,
+ "completions/mean_length": 1154.46669921875,
+ "completions/mean_terminated_length": 950.5800048828125,
+ "completions/min_length": 652.4,
+ "completions/min_terminated_length": 652.4,
+ "entropy": 0.18063361259798208,
+ "epoch": 0.07832422586520947,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.02516000345349312,
+ "kl": 0.0003309295129535409,
+ "learning_rate": 4.852377300613496e-07,
+ "loss": -0.003862759843468666,
+ "num_tokens": 1418892.0,
+ "reward": 1.766666793823242,
+ "reward_std": 0.897074180841446,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.33333333432674406,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.3333333648741245,
+ "rewards/multi_component_reward/std": 0.5403176128864289,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.24494898319244385,
+ "sampling/importance_sampling_ratio/max": 0.04162683514878154,
+ "sampling/importance_sampling_ratio/mean": 0.011416873242706061,
+ "sampling/importance_sampling_ratio/min": 0.001456704799248534,
+ "sampling/sampling_logp_difference/max": 0.3141038179397583,
+ "sampling/sampling_logp_difference/mean": 0.011231625638902187,
+ "step": 215,
+ "step_time": 40.99782377649099
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1087.4,
+ "completions/max_terminated_length": 1005.0,
+ "completions/mean_length": 687.36669921875,
+ "completions/mean_terminated_length": 652.033349609375,
+ "completions/min_length": 441.2,
+ "completions/min_terminated_length": 441.2,
+ "entropy": 0.20975149522225062,
+ "epoch": 0.08014571948998178,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.026713233441114426,
+ "kl": 0.0004683414850054154,
+ "learning_rate": 4.842791411042944e-07,
+ "loss": -0.0019101161509752274,
+ "num_tokens": 1442585.0,
+ "reward": 1.9400001287460327,
+ "reward_std": 1.0250610709190369,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.4,
+ "rewards/correctness/std": 0.45777305364608767,
+ "rewards/multi_component_reward/mean": 0.5066666960716247,
+ "rewards/multi_component_reward/std": 0.49611697196960447,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.11259435554966331,
+ "sampling/importance_sampling_ratio/mean": 0.04658988020382822,
+ "sampling/importance_sampling_ratio/min": 0.0064018826728026255,
+ "sampling/sampling_logp_difference/max": 0.35799312591552734,
+ "sampling/sampling_logp_difference/mean": 0.013789117708802224,
+ "step": 220,
+ "step_time": 25.938872035220264
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1644.6,
+ "completions/max_terminated_length": 1458.8,
+ "completions/mean_length": 1090.4333618164062,
+ "completions/mean_terminated_length": 1024.5933349609375,
+ "completions/min_length": 586.4,
+ "completions/min_terminated_length": 586.4,
+ "entropy": 0.15132917563120524,
+ "epoch": 0.08196721311475409,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.015617283061146736,
+ "kl": 0.00032217044063145297,
+ "learning_rate": 4.833205521472393e-07,
+ "loss": 0.005397457629442215,
+ "num_tokens": 1478730.0,
+ "reward": 1.2733334183692933,
+ "reward_std": 0.49739299416542054,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.10000000298023223,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.14000000953674316,
+ "rewards/multi_component_reward/std": 0.3222720980644226,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.09620626708492637,
+ "sampling/importance_sampling_ratio/mean": 0.030177546793129294,
+ "sampling/importance_sampling_ratio/min": 0.0006849504668480222,
+ "sampling/sampling_logp_difference/max": 0.35564944744110105,
+ "sampling/sampling_logp_difference/mean": 0.009922309406101703,
+ "step": 225,
+ "step_time": 38.19373752269894
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.26666667461395266,
+ "completions/max_length": 1435.0,
+ "completions/max_terminated_length": 1110.6,
+ "completions/mean_length": 1050.7667114257813,
+ "completions/mean_terminated_length": 875.8666809082031,
+ "completions/min_length": 714.8,
+ "completions/min_terminated_length": 714.8,
+ "entropy": 0.20316853250066438,
+ "epoch": 0.08378870673952642,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.021429063752293587,
+ "kl": 0.00040220142885421715,
+ "learning_rate": 4.82361963190184e-07,
+ "loss": -0.00029108244925737383,
+ "num_tokens": 1513091.0,
+ "reward": 1.3133334159851073,
+ "reward_std": 0.7383940577507019,
+ "rewards/boxed_rate/mean": 0.7333333343267441,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.1666666716337204,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.11333334147930145,
+ "rewards/multi_component_reward/std": 0.4417478919029236,
+ "rewards/no_answer_rate/mean": 0.2666666626930237,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.14741267394274474,
+ "sampling/importance_sampling_ratio/mean": 0.06669870759360493,
+ "sampling/importance_sampling_ratio/min": 0.0190697070662623,
+ "sampling/sampling_logp_difference/max": 0.34659633636474607,
+ "sampling/sampling_logp_difference/mean": 0.013104908354580402,
+ "step": 230,
+ "step_time": 33.9486582595855
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1370.2,
+ "completions/max_terminated_length": 1313.4,
+ "completions/mean_length": 1031.9666870117187,
+ "completions/mean_terminated_length": 1006.8400268554688,
+ "completions/min_length": 716.6,
+ "completions/min_terminated_length": 716.6,
+ "entropy": 0.14171662541727226,
+ "epoch": 0.08561020036429873,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.031329505145549774,
+ "kl": 0.0002987771406575727,
+ "learning_rate": 4.814033742331288e-07,
+ "loss": -0.0023104714229702948,
+ "num_tokens": 1546912.0,
+ "reward": 1.7100001096725463,
+ "reward_std": 0.7802696824073792,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.3,
+ "rewards/correctness/std": 0.34822853803634646,
+ "rewards/multi_component_reward/mean": 0.41000002026557925,
+ "rewards/multi_component_reward/std": 0.43204120099544524,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.06215879218652844,
+ "sampling/importance_sampling_ratio/mean": 0.030630824994295834,
+ "sampling/importance_sampling_ratio/min": 0.007751126746109232,
+ "sampling/sampling_logp_difference/max": 0.3309502720832825,
+ "sampling/sampling_logp_difference/mean": 0.009595037158578634,
+ "step": 235,
+ "step_time": 33.03153369966894
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.3000000029802322,
+ "completions/max_length": 1976.2,
+ "completions/max_terminated_length": 1429.4,
+ "completions/mean_length": 1217.8667114257812,
+ "completions/mean_terminated_length": 870.3533569335938,
+ "completions/min_length": 525.4,
+ "completions/min_terminated_length": 525.4,
+ "entropy": 0.1393390517681837,
+ "epoch": 0.08743169398907104,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.05459148809313774,
+ "kl": 0.0003685045870952308,
+ "learning_rate": 4.804447852760736e-07,
+ "loss": -0.0035786613821983337,
+ "num_tokens": 1587708.0,
+ "reward": 1.52666677236557,
+ "reward_std": 0.9796114563941956,
+ "rewards/boxed_rate/mean": 0.7,
+ "rewards/boxed_rate/std": 0.404018247127533,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.3823883533477783,
+ "rewards/multi_component_reward/mean": 0.19333335161209106,
+ "rewards/multi_component_reward/std": 0.6527607858180999,
+ "rewards/no_answer_rate/mean": 0.3000000029802322,
+ "rewards/no_answer_rate/std": 0.404018247127533,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.08990185633301735,
+ "sampling/importance_sampling_ratio/mean": 0.030038297548890112,
+ "sampling/importance_sampling_ratio/min": 0.00043211893236647645,
+ "sampling/sampling_logp_difference/max": 0.38284850120544434,
+ "sampling/sampling_logp_difference/mean": 0.008784445654600858,
+ "step": 240,
+ "step_time": 45.60475273691118
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1608.4,
+ "completions/max_terminated_length": 1074.2,
+ "completions/mean_length": 920.5666870117187,
+ "completions/mean_terminated_length": 754.0433471679687,
+ "completions/min_length": 548.6,
+ "completions/min_terminated_length": 548.6,
+ "entropy": 0.13041699826717376,
+ "epoch": 0.08925318761384335,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.011656666174530983,
+ "kl": 0.00033540410368004814,
+ "learning_rate": 4.794861963190184e-07,
+ "loss": -0.004278642311692238,
+ "num_tokens": 1618097.0,
+ "reward": 1.3933334350585938,
+ "reward_std": 0.31584063470363616,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.10327955484390258,
+ "rewards/multi_component_reward/mean": 0.1266666680574417,
+ "rewards/multi_component_reward/std": 0.3364715576171875,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.13333333730697633,
+ "rewards/repetition_rate/std": 0.2665788769721985,
+ "sampling/importance_sampling_ratio/max": 0.13740080818533898,
+ "sampling/importance_sampling_ratio/mean": 0.05796990524977445,
+ "sampling/importance_sampling_ratio/min": 0.006626390311389762,
+ "sampling/sampling_logp_difference/max": 0.3156670331954956,
+ "sampling/sampling_logp_difference/mean": 0.008615392539650202,
+ "step": 245,
+ "step_time": 36.637498759664595
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1533.4,
+ "completions/max_terminated_length": 1395.2,
+ "completions/mean_length": 973.1333923339844,
+ "completions/mean_terminated_length": 939.9467224121094,
+ "completions/min_length": 659.4,
+ "completions/min_terminated_length": 659.4,
+ "entropy": 0.14471415306131044,
+ "epoch": 0.09107468123861566,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.012578152120113373,
+ "kl": 0.00044469132699305193,
+ "learning_rate": 4.785276073619632e-07,
+ "loss": 0.00015797601081430913,
+ "num_tokens": 1651371.0,
+ "reward": 2.2233334302902223,
+ "reward_std": 0.6588261000812053,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.5333333432674408,
+ "rewards/correctness/std": 0.3098386645317078,
+ "rewards/multi_component_reward/mean": 0.656666674464941,
+ "rewards/multi_component_reward/std": 0.41430723667144775,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.18325970163568855,
+ "sampling/importance_sampling_ratio/mean": 0.10900020101107658,
+ "sampling/importance_sampling_ratio/min": 0.058780076946641203,
+ "sampling/sampling_logp_difference/max": 0.28016397953033445,
+ "sampling/sampling_logp_difference/mean": 0.009201886225491762,
+ "step": 250,
+ "step_time": 36.18357318136841
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1565.8,
+ "completions/max_terminated_length": 1137.4,
+ "completions/mean_length": 961.133349609375,
+ "completions/mean_terminated_length": 823.0366821289062,
+ "completions/min_length": 616.4,
+ "completions/min_terminated_length": 616.4,
+ "entropy": 0.14495401444534461,
+ "epoch": 0.09289617486338798,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.010682695545256138,
+ "kl": 0.00035484658534793806,
+ "learning_rate": 4.775690184049079e-07,
+ "loss": 0.0014964478090405465,
+ "num_tokens": 1682425.0,
+ "reward": 1.8333334445953369,
+ "reward_std": 0.6724297642707825,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.3666666716337204,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.4333333522081375,
+ "rewards/multi_component_reward/std": 0.4199430406093597,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.10162204280495643,
+ "sampling/importance_sampling_ratio/mean": 0.033536690101027486,
+ "sampling/importance_sampling_ratio/min": 0.00426993980855085,
+ "sampling/sampling_logp_difference/max": 0.33833720684051516,
+ "sampling/sampling_logp_difference/mean": 0.009549971111118794,
+ "step": 255,
+ "step_time": 35.81306515969336
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.4000000089406967,
+ "completions/max_length": 2048.0,
+ "completions/max_terminated_length": 1168.6,
+ "completions/mean_length": 1366.4000366210937,
+ "completions/mean_terminated_length": 998.9466796875,
+ "completions/min_length": 838.8,
+ "completions/min_terminated_length": 838.8,
+ "entropy": 0.14179731508096058,
+ "epoch": 0.0947176684881603,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.05714869126677513,
+ "kl": 0.00032895591721171513,
+ "learning_rate": 4.7661042944785273e-07,
+ "loss": -0.004640129953622818,
+ "num_tokens": 1726081.0,
+ "reward": 1.4900000929832458,
+ "reward_std": 0.996655923128128,
+ "rewards/boxed_rate/mean": 0.6000000059604644,
+ "rewards/boxed_rate/std": 0.5010328412055969,
+ "rewards/correctness/mean": 0.2666666716337204,
+ "rewards/correctness/std": 0.3977532863616943,
+ "rewards/multi_component_reward/mean": 0.12333334572613239,
+ "rewards/multi_component_reward/std": 0.6989697873592376,
+ "rewards/no_answer_rate/mean": 0.4000000089406967,
+ "rewards/no_answer_rate/std": 0.5010328412055969,
+ "rewards/repetition_rate/mean": 0.1,
+ "rewards/repetition_rate/std": 0.10954451560974121,
+ "sampling/importance_sampling_ratio/max": 0.043305744789540766,
+ "sampling/importance_sampling_ratio/mean": 0.01504859896376729,
+ "sampling/importance_sampling_ratio/min": 4.04306270474383e-05,
+ "sampling/sampling_logp_difference/max": 0.3569210052490234,
+ "sampling/sampling_logp_difference/mean": 0.009369691275060177,
+ "step": 260,
+ "step_time": 46.90199479162693
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2666666716337204,
+ "completions/max_length": 1583.0,
+ "completions/max_terminated_length": 934.6,
+ "completions/mean_length": 1023.133349609375,
+ "completions/mean_terminated_length": 685.6266723632813,
+ "completions/min_length": 537.4,
+ "completions/min_terminated_length": 537.4,
+ "entropy": 0.17780156135559083,
+ "epoch": 0.0965391621129326,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.018404532223939896,
+ "kl": 0.0003682690895705794,
+ "learning_rate": 4.756518404907975e-07,
+ "loss": 0.012411002814769746,
+ "num_tokens": 1759811.0,
+ "reward": 1.3833333849906921,
+ "reward_std": 0.7943651616573334,
+ "rewards/boxed_rate/mean": 0.7333333313465118,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.15000001043081285,
+ "rewards/multi_component_reward/std": 0.5146282583475112,
+ "rewards/no_answer_rate/mean": 0.2666666716337204,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.1577057947986759,
+ "sampling/importance_sampling_ratio/mean": 0.09503802299732342,
+ "sampling/importance_sampling_ratio/min": 0.0009112325180752826,
+ "sampling/sampling_logp_difference/max": 0.3926606416702271,
+ "sampling/sampling_logp_difference/mean": 0.012097407877445222,
+ "step": 265,
+ "step_time": 36.53071039505303
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1533.2,
+ "completions/max_terminated_length": 932.6,
+ "completions/mean_length": 900.7333618164063,
+ "completions/mean_terminated_length": 726.6700073242188,
+ "completions/min_length": 482.4,
+ "completions/min_terminated_length": 482.4,
+ "entropy": 0.18941813930869103,
+ "epoch": 0.09836065573770492,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.1255546510219574,
+ "kl": 0.0004568617567808057,
+ "learning_rate": 4.746932515337423e-07,
+ "loss": -0.006510256975889206,
+ "num_tokens": 1790271.0,
+ "reward": 1.7666667699813843,
+ "reward_std": 1.1080477476119994,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.33333333730697634,
+ "rewards/correctness/std": 0.48566790819168093,
+ "rewards/multi_component_reward/mean": 0.33333334922790525,
+ "rewards/multi_component_reward/std": 0.6733725190162658,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.18492921590805053,
+ "sampling/importance_sampling_ratio/max": 0.0878407845273614,
+ "sampling/importance_sampling_ratio/mean": 0.026912105677183717,
+ "sampling/importance_sampling_ratio/min": 0.0006732676178655298,
+ "sampling/sampling_logp_difference/max": 0.33626770973205566,
+ "sampling/sampling_logp_difference/mean": 0.012423686124384404,
+ "step": 270,
+ "step_time": 35.3790395591408
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1538.0,
+ "completions/max_terminated_length": 1314.2,
+ "completions/mean_length": 987.2333618164063,
+ "completions/mean_terminated_length": 886.1933471679688,
+ "completions/min_length": 572.4,
+ "completions/min_terminated_length": 572.4,
+ "entropy": 0.17079306667049726,
+ "epoch": 0.10018214936247723,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.0078973937779665,
+ "kl": 0.00039174315752461554,
+ "learning_rate": 4.737346625766871e-07,
+ "loss": -0.0023402150720357893,
+ "num_tokens": 1822306.0,
+ "reward": 1.833333396911621,
+ "reward_std": 0.9216741323471069,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.3666666716337204,
+ "rewards/correctness/std": 0.404018247127533,
+ "rewards/multi_component_reward/mean": 0.43333334028720855,
+ "rewards/multi_component_reward/std": 0.5590049713850022,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.05113908690400422,
+ "sampling/importance_sampling_ratio/mean": 0.021760703669860958,
+ "sampling/importance_sampling_ratio/min": 0.008651768978314297,
+ "sampling/sampling_logp_difference/max": 0.4047112584114075,
+ "sampling/sampling_logp_difference/mean": 0.01078464426100254,
+ "step": 275,
+ "step_time": 35.81637797001749
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1824.0,
+ "completions/max_terminated_length": 1109.4,
+ "completions/mean_length": 981.833349609375,
+ "completions/mean_terminated_length": 751.3400024414062,
+ "completions/min_length": 517.2,
+ "completions/min_terminated_length": 517.2,
+ "entropy": 0.15676350990931193,
+ "epoch": 0.10200364298724955,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.01452692411839962,
+ "kl": 0.0003485525502280022,
+ "learning_rate": 4.727760736196319e-07,
+ "loss": -0.004090853407979012,
+ "num_tokens": 1855067.0,
+ "reward": 1.4933334350585938,
+ "reward_std": 0.7522169463336468,
+ "rewards/boxed_rate/mean": 0.799999988079071,
+ "rewards/boxed_rate/std": 0.35449349880218506,
+ "rewards/correctness/mean": 0.23333333432674408,
+ "rewards/correctness/std": 0.30073869228363037,
+ "rewards/multi_component_reward/mean": 0.2266666792333126,
+ "rewards/multi_component_reward/std": 0.5248861283063888,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.35449349880218506,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.12218071203678846,
+ "sampling/importance_sampling_ratio/mean": 0.04405408292077482,
+ "sampling/importance_sampling_ratio/min": 0.00501076535471764,
+ "sampling/sampling_logp_difference/max": 0.37753498554229736,
+ "sampling/sampling_logp_difference/mean": 0.009699719026684761,
+ "step": 280,
+ "step_time": 40.97467570956796
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1744.4,
+ "completions/max_terminated_length": 1173.6,
+ "completions/mean_length": 970.86669921875,
+ "completions/mean_terminated_length": 758.7533447265625,
+ "completions/min_length": 479.4,
+ "completions/min_terminated_length": 479.4,
+ "entropy": 0.15489849572380385,
+ "epoch": 0.10382513661202186,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.09483744204044342,
+ "kl": 0.000336567038417949,
+ "learning_rate": 4.718174846625766e-07,
+ "loss": -0.004700837284326553,
+ "num_tokens": 1886869.0,
+ "reward": 1.7000001311302184,
+ "reward_std": 0.8241950333118438,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.33333333730697634,
+ "rewards/correctness/std": 0.32236858606338503,
+ "rewards/multi_component_reward/mean": 0.36666667461395264,
+ "rewards/multi_component_reward/std": 0.5094490587711334,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.13168737590312957,
+ "sampling/importance_sampling_ratio/mean": 0.06171476859599352,
+ "sampling/importance_sampling_ratio/min": 0.010470659041346208,
+ "sampling/sampling_logp_difference/max": 0.344563889503479,
+ "sampling/sampling_logp_difference/mean": 0.010558286216109991,
+ "step": 285,
+ "step_time": 39.89412596244365
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333333432674408,
+ "completions/max_length": 1809.2,
+ "completions/max_terminated_length": 1132.6,
+ "completions/mean_length": 1119.7000366210937,
+ "completions/mean_terminated_length": 832.9400268554688,
+ "completions/min_length": 590.4,
+ "completions/min_terminated_length": 590.4,
+ "entropy": 0.15704507902264594,
+ "epoch": 0.10564663023679417,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.023737642914056778,
+ "kl": 0.0003448050100511561,
+ "learning_rate": 4.7085889570552147e-07,
+ "loss": 0.0019955366849899294,
+ "num_tokens": 1923040.0,
+ "reward": 1.380000114440918,
+ "reward_std": 0.5739886239171028,
+ "rewards/boxed_rate/mean": 0.7666666626930236,
+ "rewards/boxed_rate/std": 0.30073869228363037,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2128240704536438,
+ "rewards/multi_component_reward/mean": 0.11333334371447563,
+ "rewards/multi_component_reward/std": 0.46365060210227965,
+ "rewards/no_answer_rate/mean": 0.23333333432674408,
+ "rewards/no_answer_rate/std": 0.30073869228363037,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.18492921590805053,
+ "sampling/importance_sampling_ratio/max": 0.04411560408771038,
+ "sampling/importance_sampling_ratio/mean": 0.01753278383985162,
+ "sampling/importance_sampling_ratio/min": 0.0009861058487468544,
+ "sampling/sampling_logp_difference/max": 0.3644953012466431,
+ "sampling/sampling_logp_difference/mean": 0.010005151480436325,
+ "step": 290,
+ "step_time": 41.05979204457253
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1592.8,
+ "completions/max_terminated_length": 1143.4,
+ "completions/mean_length": 987.1000366210938,
+ "completions/mean_terminated_length": 825.9633544921875,
+ "completions/min_length": 591.8,
+ "completions/min_terminated_length": 591.8,
+ "entropy": 0.16288827657699584,
+ "epoch": 0.10746812386156648,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.026686564087867737,
+ "kl": 0.0003618090704549104,
+ "learning_rate": 4.6990030674846625e-07,
+ "loss": -0.0018155150115489959,
+ "num_tokens": 1955647.0,
+ "reward": 1.7666668176651001,
+ "reward_std": 0.4796482801437378,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.33333333432674406,
+ "rewards/correctness/std": 0.19119417667388916,
+ "rewards/multi_component_reward/mean": 0.36666668206453323,
+ "rewards/multi_component_reward/std": 0.34831122159957884,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.17200291641056537,
+ "sampling/importance_sampling_ratio/mean": 0.09221775899641216,
+ "sampling/importance_sampling_ratio/min": 0.053504807552870126,
+ "sampling/sampling_logp_difference/max": 0.31802849769592284,
+ "sampling/sampling_logp_difference/mean": 0.010488973837345839,
+ "step": 295,
+ "step_time": 36.921972643770275
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1181.2,
+ "completions/max_terminated_length": 1068.4,
+ "completions/mean_length": 714.5666870117187,
+ "completions/mean_terminated_length": 675.57333984375,
+ "completions/min_length": 433.8,
+ "completions/min_terminated_length": 433.8,
+ "entropy": 0.20909111325939497,
+ "epoch": 0.1092896174863388,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.9235589504241943,
+ "kl": 0.0007714893268712331,
+ "learning_rate": 4.68941717791411e-07,
+ "loss": 0.0028855174779891966,
+ "num_tokens": 1980240.0,
+ "reward": 1.8966667413711549,
+ "reward_std": 1.0613920927047729,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.3666666716337204,
+ "rewards/correctness/std": 0.5072978019714356,
+ "rewards/multi_component_reward/mean": 0.4633333578705788,
+ "rewards/multi_component_reward/std": 0.5980847001075744,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.11060594655573368,
+ "sampling/importance_sampling_ratio/mean": 0.05734073922503739,
+ "sampling/importance_sampling_ratio/min": 0.024227803308775008,
+ "sampling/sampling_logp_difference/max": 0.31456995010375977,
+ "sampling/sampling_logp_difference/mean": 0.013103757984936237,
+ "step": 300,
+ "step_time": 27.952531585469842
+ },
+ {
+ "epoch": 0.1092896174863388,
+ "eval_clip_ratio/high_max": 0.0,
+ "eval_clip_ratio/high_mean": 0.0,
+ "eval_clip_ratio/low_mean": 0.0,
+ "eval_clip_ratio/low_min": 0.0,
+ "eval_clip_ratio/region_mean": 0.0,
+ "eval_completions/clipped_ratio": 0.22666667222976686,
+ "eval_completions/max_length": 1758.54,
+ "eval_completions/max_terminated_length": 1125.4,
+ "eval_completions/mean_length": 1107.770029296875,
+ "eval_completions/mean_terminated_length": 846.1800146484375,
+ "eval_completions/min_length": 606.8,
+ "eval_completions/min_terminated_length": 606.8,
+ "eval_entropy": 0.17873862124979495,
+ "eval_frac_reward_zero_std": 0.08,
+ "eval_kl": 0.0003903971218096558,
+ "eval_loss": -0.0051070391200482845,
+ "eval_num_tokens": 1980240.0,
+ "eval_reward": 1.5980000698566437,
+ "eval_reward_std": 0.782314371317625,
+ "eval_rewards/boxed_rate/mean": 0.7766666629910469,
+ "eval_rewards/boxed_rate/std": 0.33167909026145936,
+ "eval_rewards/correctness/mean": 0.28666666984558103,
+ "eval_rewards/correctness/std": 0.30271870851516725,
+ "eval_rewards/multi_component_reward/mean": 0.26800001293420794,
+ "eval_rewards/multi_component_reward/std": 0.5066750717163085,
+ "eval_rewards/no_answer_rate/mean": 0.22333333760499954,
+ "eval_rewards/no_answer_rate/std": 0.33167909026145936,
+ "eval_rewards/repetition_rate/mean": 0.04333333432674408,
+ "eval_rewards/repetition_rate/std": 0.08060015916824341,
+ "eval_runtime": 1753.4069,
+ "eval_samples_per_second": 0.029,
+ "eval_sampling/importance_sampling_ratio/max": 0.0897229278186569,
+ "eval_sampling/importance_sampling_ratio/mean": 0.035299012192626836,
+ "eval_sampling/importance_sampling_ratio/min": 0.0025020323396462286,
+ "eval_sampling/sampling_logp_difference/max": 0.5019185048341751,
+ "eval_sampling/sampling_logp_difference/mean": 0.012026464603841304,
+ "eval_steps_per_second": 0.005,
+ "step": 300
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1,
+ "completions/max_length": 1605.6,
+ "completions/max_terminated_length": 1384.6,
+ "completions/mean_length": 1010.7000366210938,
+ "completions/mean_terminated_length": 876.26669921875,
+ "completions/min_length": 573.8,
+ "completions/min_terminated_length": 573.8,
+ "entropy": 0.1340769293407599,
+ "epoch": 0.1111111111111111,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.034374091774225235,
+ "kl": 0.000321884588629473,
+ "learning_rate": 4.6798312883435583e-07,
+ "loss": -0.014604152739048004,
+ "num_tokens": 2014467.0,
+ "reward": 2.043333411216736,
+ "reward_std": 0.8047043204307556,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.10954451560974121,
+ "rewards/correctness/mean": 0.4666666716337204,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.5433333471417428,
+ "rewards/multi_component_reward/std": 0.4605102062225342,
+ "rewards/no_answer_rate/mean": 0.1,
+ "rewards/no_answer_rate/std": 0.10954451560974121,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.12428842782974243,
+ "sampling/importance_sampling_ratio/mean": 0.046624291501939294,
+ "sampling/importance_sampling_ratio/min": 0.0006754371514054114,
+ "sampling/sampling_logp_difference/max": 0.32104220390319826,
+ "sampling/sampling_logp_difference/mean": 0.008986939676105976,
+ "step": 305,
+ "step_time": 36.810268136300145
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1557.0,
+ "completions/max_terminated_length": 1291.8,
+ "completions/mean_length": 1068.033349609375,
+ "completions/mean_terminated_length": 916.2,
+ "completions/min_length": 631.6,
+ "completions/min_terminated_length": 631.6,
+ "entropy": 0.15544118496278922,
+ "epoch": 0.11293260473588343,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.01212399173527956,
+ "kl": 0.00047503276291536165,
+ "learning_rate": 4.670245398773006e-07,
+ "loss": -8.383383974432945e-05,
+ "num_tokens": 2049616.0,
+ "reward": 1.6533334255218506,
+ "reward_std": 1.070458745956421,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2128240704536438,
+ "rewards/correctness/mean": 0.3000000059604645,
+ "rewards/correctness/std": 0.4794029474258423,
+ "rewards/multi_component_reward/mean": 0.3200000122189522,
+ "rewards/multi_component_reward/std": 0.6258516371250152,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2128240704536438,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.08939264337532223,
+ "sampling/importance_sampling_ratio/mean": 0.030286337621510027,
+ "sampling/importance_sampling_ratio/min": 0.0005470938450967466,
+ "sampling/sampling_logp_difference/max": 0.34738160371780397,
+ "sampling/sampling_logp_difference/mean": 0.010354180075228214,
+ "step": 310,
+ "step_time": 36.12143337074667
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.3333333432674408,
+ "completions/max_length": 1634.2,
+ "completions/max_terminated_length": 1091.4,
+ "completions/mean_length": 1218.6333618164062,
+ "completions/mean_terminated_length": 858.4666870117187,
+ "completions/min_length": 677.6,
+ "completions/min_terminated_length": 677.6,
+ "entropy": 0.17260866140325865,
+ "epoch": 0.11475409836065574,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.05682944133877754,
+ "kl": 0.0005160604792763479,
+ "learning_rate": 4.6606595092024536e-07,
+ "loss": -0.006619427353143692,
+ "num_tokens": 2088659.0,
+ "reward": 2.090000092983246,
+ "reward_std": 1.1096696853637695,
+ "rewards/boxed_rate/mean": 0.6666666716337204,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.566666665673256,
+ "rewards/correctness/std": 0.451508092880249,
+ "rewards/multi_component_reward/mean": 0.523333391547203,
+ "rewards/multi_component_reward/std": 0.6607022762298584,
+ "rewards/no_answer_rate/mean": 0.33333333134651183,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.05110827875323594,
+ "sampling/importance_sampling_ratio/mean": 0.018865692079998553,
+ "sampling/importance_sampling_ratio/min": 0.0009169460441439363,
+ "sampling/sampling_logp_difference/max": 0.3731674671173096,
+ "sampling/sampling_logp_difference/mean": 0.01166527420282364,
+ "step": 315,
+ "step_time": 38.15295873656869
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1484.2,
+ "completions/max_terminated_length": 1039.2,
+ "completions/mean_length": 931.7000244140625,
+ "completions/mean_terminated_length": 790.5233520507812,
+ "completions/min_length": 563.4,
+ "completions/min_terminated_length": 563.4,
+ "entropy": 0.143128818521897,
+ "epoch": 0.11657559198542805,
+ "frac_reward_zero_std": 0.6,
+ "grad_norm": 0.005422326270490885,
+ "kl": 0.0004371852917150439,
+ "learning_rate": 4.6510736196319015e-07,
+ "loss": 0.004143273830413819,
+ "num_tokens": 2119250.0,
+ "reward": 1.5300001502037048,
+ "reward_std": 0.24947773814201354,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.23333333432674408,
+ "rewards/correctness/std": 0.08164966106414795,
+ "rewards/multi_component_reward/mean": 0.29666668176651,
+ "rewards/multi_component_reward/std": 0.17254199385643004,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.12071083066985011,
+ "sampling/importance_sampling_ratio/mean": 0.05540300235152244,
+ "sampling/importance_sampling_ratio/min": 0.013269457093102949,
+ "sampling/sampling_logp_difference/max": 0.36359539031982424,
+ "sampling/sampling_logp_difference/mean": 0.009230617992579937,
+ "step": 320,
+ "step_time": 34.49489349294454
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1798.0,
+ "completions/max_terminated_length": 1537.6,
+ "completions/mean_length": 1144.8333618164063,
+ "completions/mean_terminated_length": 965.6733520507812,
+ "completions/min_length": 634.0,
+ "completions/min_terminated_length": 634.0,
+ "entropy": 0.18388984675208728,
+ "epoch": 0.11839708561020036,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.14283621311187744,
+ "kl": 0.0006492346903542057,
+ "learning_rate": 4.64148773006135e-07,
+ "loss": -0.0025676295161247255,
+ "num_tokens": 2157267.0,
+ "reward": 1.630000066757202,
+ "reward_std": 0.89271479845047,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.3000000059604645,
+ "rewards/correctness/std": 0.3761233925819397,
+ "rewards/multi_component_reward/mean": 0.3300000101327896,
+ "rewards/multi_component_reward/std": 0.5265865564346314,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.04229360092431307,
+ "sampling/importance_sampling_ratio/mean": 0.019075372768566013,
+ "sampling/importance_sampling_ratio/min": 8.805350876674933e-05,
+ "sampling/sampling_logp_difference/max": 0.3247460603713989,
+ "sampling/sampling_logp_difference/mean": 0.01177409002557397,
+ "step": 325,
+ "step_time": 41.04225417692214
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.0,
+ "completions/max_length": 823.4,
+ "completions/max_terminated_length": 823.4,
+ "completions/mean_length": 544.1333435058593,
+ "completions/mean_terminated_length": 544.1333435058593,
+ "completions/min_length": 366.0,
+ "completions/min_terminated_length": 366.0,
+ "entropy": 0.22606497158606847,
+ "epoch": 0.12021857923497267,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.9119893908500671,
+ "kl": 0.0007021349136872838,
+ "learning_rate": 4.631901840490797e-07,
+ "loss": -0.006102682277560234,
+ "num_tokens": 2175259.0,
+ "reward": 1.8700001001358033,
+ "reward_std": 0.8484382510185242,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.3666666716337204,
+ "rewards/correctness/std": 0.404018247127533,
+ "rewards/multi_component_reward/mean": 0.503333343565464,
+ "rewards/multi_component_reward/std": 0.44442008137702943,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.13480582600459456,
+ "sampling/importance_sampling_ratio/mean": 0.07863845487590879,
+ "sampling/importance_sampling_ratio/min": 0.01574344330747408,
+ "sampling/sampling_logp_difference/max": 0.3204054594039917,
+ "sampling/sampling_logp_difference/mean": 0.014066471531987191,
+ "step": 330,
+ "step_time": 20.059878361225127
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1395.2,
+ "completions/max_terminated_length": 910.6,
+ "completions/mean_length": 742.1333679199219,
+ "completions/mean_terminated_length": 640.480029296875,
+ "completions/min_length": 509.6,
+ "completions/min_terminated_length": 509.6,
+ "entropy": 0.2093804210424423,
+ "epoch": 0.122040072859745,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.028569765388965607,
+ "kl": 0.0006724427085525046,
+ "learning_rate": 4.622315950920245e-07,
+ "loss": -0.0013746816664934158,
+ "num_tokens": 2200067.0,
+ "reward": 1.6200001239776611,
+ "reward_std": 0.8776960492134094,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.3823883533477783,
+ "rewards/multi_component_reward/mean": 0.3533333420753479,
+ "rewards/multi_component_reward/std": 0.49811467826366423,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.07016745954751968,
+ "sampling/importance_sampling_ratio/mean": 0.027400266751646994,
+ "sampling/importance_sampling_ratio/min": 0.0014501588469148616,
+ "sampling/sampling_logp_difference/max": 0.36046857833862306,
+ "sampling/sampling_logp_difference/mean": 0.014280413649976254,
+ "step": 335,
+ "step_time": 31.59101020414382
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000596046447,
+ "completions/max_length": 1939.2,
+ "completions/max_terminated_length": 988.8,
+ "completions/mean_length": 976.86669921875,
+ "completions/mean_terminated_length": 681.2466857910156,
+ "completions/min_length": 446.6,
+ "completions/min_terminated_length": 446.6,
+ "entropy": 0.2083836982647578,
+ "epoch": 0.12386156648451731,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.06303755939006805,
+ "kl": 0.0005712434819239812,
+ "learning_rate": 4.612730061349693e-07,
+ "loss": -0.0044234395027160645,
+ "num_tokens": 2232145.0,
+ "reward": 1.74333336353302,
+ "reward_std": 0.9833520889282227,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.33333334028720857,
+ "rewards/correctness/std": 0.3977532863616943,
+ "rewards/multi_component_reward/mean": 0.3433333486318588,
+ "rewards/multi_component_reward/std": 0.6024900764226914,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.1221102561801672,
+ "sampling/importance_sampling_ratio/mean": 0.029469438176602126,
+ "sampling/importance_sampling_ratio/min": 5.1707089343915565e-05,
+ "sampling/sampling_logp_difference/max": 0.3329290866851807,
+ "sampling/sampling_logp_difference/mean": 0.012941631115972995,
+ "step": 340,
+ "step_time": 43.22656154055149
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1,
+ "completions/max_length": 1163.2,
+ "completions/max_terminated_length": 1144.4,
+ "completions/mean_length": 778.86669921875,
+ "completions/mean_terminated_length": 688.9000244140625,
+ "completions/min_length": 431.8,
+ "completions/min_terminated_length": 431.8,
+ "entropy": 0.1657411351799965,
+ "epoch": 0.12568306010928962,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.1018831878900528,
+ "kl": 0.0005899470415897667,
+ "learning_rate": 4.603144171779141e-07,
+ "loss": -0.006323814392089844,
+ "num_tokens": 2259027.0,
+ "reward": 1.5300001144409179,
+ "reward_std": 0.9171831011772156,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.10954451560974121,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.4298781991004944,
+ "rewards/multi_component_reward/mean": 0.23000000715255736,
+ "rewards/multi_component_reward/std": 0.5341363847255707,
+ "rewards/no_answer_rate/mean": 0.1,
+ "rewards/no_answer_rate/std": 0.10954451560974121,
+ "rewards/repetition_rate/mean": 0.1,
+ "rewards/repetition_rate/std": 0.10954451560974121,
+ "sampling/importance_sampling_ratio/max": 0.12837026529014112,
+ "sampling/importance_sampling_ratio/mean": 0.05263339746743441,
+ "sampling/importance_sampling_ratio/min": 0.0009968833521444183,
+ "sampling/sampling_logp_difference/max": 0.31432443857192993,
+ "sampling/sampling_logp_difference/mean": 0.011136513762176036,
+ "step": 345,
+ "step_time": 27.575114846229553
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1772.4,
+ "completions/max_terminated_length": 1112.4,
+ "completions/mean_length": 983.0333618164062,
+ "completions/mean_terminated_length": 776.57333984375,
+ "completions/min_length": 516.6,
+ "completions/min_terminated_length": 516.6,
+ "entropy": 0.18533113847176233,
+ "epoch": 0.12750455373406194,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.1743193119764328,
+ "kl": 0.0007398304010469777,
+ "learning_rate": 4.593558282208589e-07,
+ "loss": 0.0035149652510881426,
+ "num_tokens": 2292724.0,
+ "reward": 1.5366667270660401,
+ "reward_std": 0.7906980156898499,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.34822853803634646,
+ "rewards/multi_component_reward/mean": 0.2366666778922081,
+ "rewards/multi_component_reward/std": 0.52751624584198,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.16946832239627838,
+ "sampling/importance_sampling_ratio/mean": 0.04226875379681587,
+ "sampling/importance_sampling_ratio/min": 0.00029637051690457807,
+ "sampling/sampling_logp_difference/max": 0.34900493621826173,
+ "sampling/sampling_logp_difference/mean": 0.011652881279587746,
+ "step": 350,
+ "step_time": 40.440953285992144
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1092.8,
+ "completions/max_terminated_length": 992.2,
+ "completions/mean_length": 720.4666931152344,
+ "completions/mean_terminated_length": 686.5133483886718,
+ "completions/min_length": 467.2,
+ "completions/min_terminated_length": 467.2,
+ "entropy": 0.1794678675631682,
+ "epoch": 0.12932604735883424,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.020413219928741455,
+ "kl": 0.0006120980954923046,
+ "learning_rate": 4.583972392638036e-07,
+ "loss": -0.0014667985960841179,
+ "num_tokens": 2316852.0,
+ "reward": 2.060000109672546,
+ "reward_std": 0.7259637594223023,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.46666666865348816,
+ "rewards/correctness/std": 0.32236858606338503,
+ "rewards/multi_component_reward/mean": 0.59333336353302,
+ "rewards/multi_component_reward/std": 0.4035952419042587,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.14755254588089883,
+ "sampling/importance_sampling_ratio/mean": 0.07640535607933999,
+ "sampling/importance_sampling_ratio/min": 0.03205070712427443,
+ "sampling/sampling_logp_difference/max": 0.3078671216964722,
+ "sampling/sampling_logp_difference/mean": 0.011428364552557468,
+ "step": 355,
+ "step_time": 26.221613752283154
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1573.4,
+ "completions/max_terminated_length": 1007.8,
+ "completions/mean_length": 839.6000122070312,
+ "completions/mean_terminated_length": 670.27333984375,
+ "completions/min_length": 423.0,
+ "completions/min_terminated_length": 423.0,
+ "entropy": 0.2592687545965115,
+ "epoch": 0.13114754098360656,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.21311414241790771,
+ "kl": 0.0006711701275586772,
+ "learning_rate": 4.5743865030674845e-07,
+ "loss": -0.0009625215083360672,
+ "num_tokens": 2345424.0,
+ "reward": 1.6000000953674316,
+ "reward_std": 0.7632826209068299,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.2666666656732559,
+ "rewards/correctness/std": 0.3265986442565918,
+ "rewards/multi_component_reward/mean": 0.33333335518836976,
+ "rewards/multi_component_reward/std": 0.44263782203197477,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.06810822673141956,
+ "sampling/importance_sampling_ratio/mean": 0.026497452543117105,
+ "sampling/importance_sampling_ratio/min": 0.0005043454068991798,
+ "sampling/sampling_logp_difference/max": 0.46866347789764407,
+ "sampling/sampling_logp_difference/mean": 0.01661387998610735,
+ "step": 360,
+ "step_time": 35.86930139716715
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333334028720856,
+ "completions/max_length": 1906.6,
+ "completions/max_terminated_length": 1405.8,
+ "completions/mean_length": 1317.56669921875,
+ "completions/mean_terminated_length": 1097.8466796875,
+ "completions/min_length": 801.0,
+ "completions/min_terminated_length": 801.0,
+ "entropy": 0.11761416780451933,
+ "epoch": 0.13296903460837886,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.015462023206055164,
+ "kl": 0.0003710240945414019,
+ "learning_rate": 4.5648006134969324e-07,
+ "loss": -0.007586926966905594,
+ "num_tokens": 2388119.0,
+ "reward": 1.7300000667572022,
+ "reward_std": 1.0704036235809327,
+ "rewards/boxed_rate/mean": 0.7666666686534882,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.33333333730697634,
+ "rewards/correctness/std": 0.48566790819168093,
+ "rewards/multi_component_reward/mean": 0.2966666899621487,
+ "rewards/multi_component_reward/std": 0.6780295848846436,
+ "rewards/no_answer_rate/mean": 0.23333334028720856,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.24494898319244385,
+ "sampling/importance_sampling_ratio/max": 0.04545931427273899,
+ "sampling/importance_sampling_ratio/mean": 0.02320328297209926,
+ "sampling/importance_sampling_ratio/min": 0.001607318179527617,
+ "sampling/sampling_logp_difference/max": 0.35903769731521606,
+ "sampling/sampling_logp_difference/mean": 0.007444971334189176,
+ "step": 365,
+ "step_time": 43.916714335232975
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1418.0,
+ "completions/max_terminated_length": 1135.4,
+ "completions/mean_length": 861.5666931152343,
+ "completions/mean_terminated_length": 710.4833435058594,
+ "completions/min_length": 346.6,
+ "completions/min_terminated_length": 346.6,
+ "entropy": 0.22315610001484554,
+ "epoch": 0.13479052823315119,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 1.7706270217895508,
+ "kl": 0.0008742218895349651,
+ "learning_rate": 4.55521472392638e-07,
+ "loss": 0.003686445206403732,
+ "num_tokens": 2416660.0,
+ "reward": 1.5800000548362731,
+ "reward_std": 0.8671060860157013,
+ "rewards/boxed_rate/mean": 0.8666666746139526,
+ "rewards/boxed_rate/std": 0.20655910968780516,
+ "rewards/correctness/mean": 0.26666667461395266,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.31333333253860474,
+ "rewards/multi_component_reward/std": 0.5032015323638916,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.20655910968780516,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.20161812007427216,
+ "sampling/importance_sampling_ratio/mean": 0.09829787239432335,
+ "sampling/importance_sampling_ratio/min": 0.056043115374653285,
+ "sampling/sampling_logp_difference/max": 0.35296247005462644,
+ "sampling/sampling_logp_difference/mean": 0.014671148918569087,
+ "step": 370,
+ "step_time": 33.350730242021385
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333432674407,
+ "completions/max_length": 1741.8,
+ "completions/max_terminated_length": 1278.2,
+ "completions/mean_length": 958.4000366210937,
+ "completions/mean_terminated_length": 774.8600219726562,
+ "completions/min_length": 536.0,
+ "completions/min_terminated_length": 536.0,
+ "entropy": 0.19409313425421715,
+ "epoch": 0.1366120218579235,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.10258854180574417,
+ "kl": 0.0005339390016160905,
+ "learning_rate": 4.545628834355828e-07,
+ "loss": 0.0006846859585493803,
+ "num_tokens": 2448526.0,
+ "reward": 1.6966667890548706,
+ "reward_std": 1.038489294052124,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.3000000059604645,
+ "rewards/correctness/std": 0.4794029474258423,
+ "rewards/multi_component_reward/mean": 0.3300000175833702,
+ "rewards/multi_component_reward/std": 0.6378163933753968,
+ "rewards/no_answer_rate/mean": 0.13333333432674407,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.023186815530061723,
+ "sampling/importance_sampling_ratio/mean": 0.013102744333446026,
+ "sampling/importance_sampling_ratio/min": 0.00010366742491344506,
+ "sampling/sampling_logp_difference/max": 0.3774606704711914,
+ "sampling/sampling_logp_difference/mean": 0.011916628293693065,
+ "step": 375,
+ "step_time": 30.433830817975103
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333334028720856,
+ "completions/max_length": 1764.6,
+ "completions/max_terminated_length": 1185.0,
+ "completions/mean_length": 1108.233349609375,
+ "completions/mean_terminated_length": 801.3400024414062,
+ "completions/min_length": 541.8,
+ "completions/min_terminated_length": 541.8,
+ "entropy": 0.25079266702135405,
+ "epoch": 0.1384335154826958,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.015607405453920364,
+ "kl": 0.000638273500953801,
+ "learning_rate": 4.536042944785276e-07,
+ "loss": -0.004855351522564888,
+ "num_tokens": 2484623.0,
+ "reward": 1.6300000548362732,
+ "reward_std": 0.6760274231433868,
+ "rewards/boxed_rate/mean": 0.799999988079071,
+ "rewards/boxed_rate/std": 0.35449349880218506,
+ "rewards/correctness/mean": 0.2666666656732559,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.23000002317130566,
+ "rewards/multi_component_reward/std": 0.4806109189987183,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.35449349880218506,
+ "rewards/repetition_rate/mean": 0.13333333432674407,
+ "rewards/repetition_rate/std": 0.19119417667388916,
+ "sampling/importance_sampling_ratio/max": 0.030251298751682042,
+ "sampling/importance_sampling_ratio/mean": 0.01690196809358895,
+ "sampling/importance_sampling_ratio/min": 0.0027151843596084375,
+ "sampling/sampling_logp_difference/max": 0.3498866558074951,
+ "sampling/sampling_logp_difference/mean": 0.016159016452729703,
+ "step": 380,
+ "step_time": 15.216538714803756
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000596046447,
+ "completions/max_length": 1980.6,
+ "completions/max_terminated_length": 1102.2,
+ "completions/mean_length": 1102.2000122070312,
+ "completions/mean_terminated_length": 820.3700073242187,
+ "completions/min_length": 554.4,
+ "completions/min_terminated_length": 554.4,
+ "entropy": 0.21501451556881268,
+ "epoch": 0.14025500910746813,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.039810437709093094,
+ "kl": 0.0005428103109200795,
+ "learning_rate": 4.5264570552147234e-07,
+ "loss": -0.001014799065887928,
+ "num_tokens": 2521715.0,
+ "reward": 1.33000009059906,
+ "reward_std": 0.713744330406189,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.36985843181610106,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.16333335041999816,
+ "rewards/multi_component_reward/std": 0.4538270115852356,
+ "rewards/no_answer_rate/mean": 0.20000000596046447,
+ "rewards/no_answer_rate/std": 0.36985843181610106,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.050251276697963476,
+ "sampling/importance_sampling_ratio/mean": 0.014141232566908002,
+ "sampling/importance_sampling_ratio/min": 1.2361302467010191e-05,
+ "sampling/sampling_logp_difference/max": 0.30763993263244627,
+ "sampling/sampling_logp_difference/mean": 0.014410833641886712,
+ "step": 385,
+ "step_time": 16.672282354161144
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.33333333730697634,
+ "completions/max_length": 1734.4,
+ "completions/max_terminated_length": 1527.6,
+ "completions/mean_length": 1290.500048828125,
+ "completions/mean_terminated_length": 1057.36669921875,
+ "completions/min_length": 721.2,
+ "completions/min_terminated_length": 721.2,
+ "entropy": 0.17866969083746273,
+ "epoch": 0.14207650273224043,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.009094689041376114,
+ "kl": 0.0004922606613642226,
+ "learning_rate": 4.5168711656441713e-07,
+ "loss": -0.0016124360263347626,
+ "num_tokens": 2563196.0,
+ "reward": 1.5066667079925538,
+ "reward_std": 0.9714008927345276,
+ "rewards/boxed_rate/mean": 0.6666666686534881,
+ "rewards/boxed_rate/std": 0.32236858606338503,
+ "rewards/correctness/mean": 0.26666667461395266,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.17333333306014537,
+ "rewards/multi_component_reward/std": 0.564999234676361,
+ "rewards/no_answer_rate/mean": 0.33333333730697634,
+ "rewards/no_answer_rate/std": 0.32236858606338503,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.09146881643682718,
+ "sampling/importance_sampling_ratio/mean": 0.027744953380897643,
+ "sampling/importance_sampling_ratio/min": 0.0008299978789871564,
+ "sampling/sampling_logp_difference/max": 0.44667046070098876,
+ "sampling/sampling_logp_difference/mean": 0.011561266146600246,
+ "step": 390,
+ "step_time": 15.62863542791456
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1713.0,
+ "completions/max_terminated_length": 1077.4,
+ "completions/mean_length": 995.9333801269531,
+ "completions/mean_terminated_length": 786.106689453125,
+ "completions/min_length": 546.4,
+ "completions/min_terminated_length": 546.4,
+ "entropy": 0.1970411961277326,
+ "epoch": 0.14389799635701275,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.018740270286798477,
+ "kl": 0.0007653967642302935,
+ "learning_rate": 4.50728527607362e-07,
+ "loss": 0.005758083239197731,
+ "num_tokens": 2595858.0,
+ "reward": 1.4666666984558105,
+ "reward_std": 0.6528747193515301,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.20000000223517417,
+ "rewards/multi_component_reward/std": 0.46670507490634916,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.09698612228967249,
+ "sampling/importance_sampling_ratio/mean": 0.030241910181939603,
+ "sampling/importance_sampling_ratio/min": 0.0021489314406773754,
+ "sampling/sampling_logp_difference/max": 0.34886856079101564,
+ "sampling/sampling_logp_difference/mean": 0.012738966010510921,
+ "step": 395,
+ "step_time": 14.849660068005324
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1575.0,
+ "completions/max_terminated_length": 948.8,
+ "completions/mean_length": 935.7000366210938,
+ "completions/mean_terminated_length": 702.1866821289062,
+ "completions/min_length": 482.0,
+ "completions/min_terminated_length": 482.0,
+ "entropy": 0.2041997093707323,
+ "epoch": 0.14571948998178508,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.10930619388818741,
+ "kl": 0.0005768983881959381,
+ "learning_rate": 4.497699386503067e-07,
+ "loss": -0.004532752931118012,
+ "num_tokens": 2626437.0,
+ "reward": 1.7466667890548706,
+ "reward_std": 0.8040140867233276,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.33333333134651183,
+ "rewards/correctness/std": 0.35449349880218506,
+ "rewards/multi_component_reward/mean": 0.34666669368743896,
+ "rewards/multi_component_reward/std": 0.5015990376472473,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.09340296909213067,
+ "sampling/importance_sampling_ratio/mean": 0.03166428431868553,
+ "sampling/importance_sampling_ratio/min": 0.007079391362589009,
+ "sampling/sampling_logp_difference/max": 0.4040722370147705,
+ "sampling/sampling_logp_difference/mean": 0.013007087912410498,
+ "step": 400,
+ "step_time": 13.86912056710571
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1673.6,
+ "completions/max_terminated_length": 1493.4,
+ "completions/mean_length": 1166.2000366210937,
+ "completions/mean_terminated_length": 1052.9500122070312,
+ "completions/min_length": 725.0,
+ "completions/min_terminated_length": 725.0,
+ "entropy": 0.1420179379483064,
+ "epoch": 0.14754098360655737,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.021460620686411858,
+ "kl": 0.0005037313070109425,
+ "learning_rate": 4.488113496932515e-07,
+ "loss": -0.0034239646047353745,
+ "num_tokens": 2665191.0,
+ "reward": 1.6500000953674316,
+ "reward_std": 0.5699952363967895,
+ "rewards/boxed_rate/mean": 0.8666666746139526,
+ "rewards/boxed_rate/std": 0.20655910968780516,
+ "rewards/correctness/mean": 0.2666666656732559,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.2833333633840084,
+ "rewards/multi_component_reward/std": 0.40620826482772826,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.20655910968780516,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.18492921590805053,
+ "sampling/importance_sampling_ratio/max": 0.047187263565137985,
+ "sampling/importance_sampling_ratio/mean": 0.016343344282358886,
+ "sampling/importance_sampling_ratio/min": 0.0005224805446630398,
+ "sampling/sampling_logp_difference/max": 0.33698827028274536,
+ "sampling/sampling_logp_difference/mean": 0.009091775678098202,
+ "step": 405,
+ "step_time": 14.932769652456045
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2333333373069763,
+ "completions/max_length": 1674.6,
+ "completions/max_terminated_length": 1293.0,
+ "completions/mean_length": 1194.5333679199218,
+ "completions/mean_terminated_length": 945.5333374023437,
+ "completions/min_length": 665.2,
+ "completions/min_terminated_length": 665.2,
+ "entropy": 0.19754986638824146,
+ "epoch": 0.1493624772313297,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.12479820102453232,
+ "kl": 0.000494376394393233,
+ "learning_rate": 4.4785276073619634e-07,
+ "loss": -0.0048382353037595745,
+ "num_tokens": 2707231.0,
+ "reward": 1.240000104904175,
+ "reward_std": 0.5429712414741517,
+ "rewards/boxed_rate/mean": 0.7666666746139527,
+ "rewards/boxed_rate/std": 0.3161036252975464,
+ "rewards/correctness/mean": 0.13333333432674407,
+ "rewards/correctness/std": 0.19119417667388916,
+ "rewards/multi_component_reward/mean": 0.10666667819023132,
+ "rewards/multi_component_reward/std": 0.35998966693878176,
+ "rewards/no_answer_rate/mean": 0.2333333373069763,
+ "rewards/no_answer_rate/std": 0.3161036252975464,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.10651108641177416,
+ "sampling/importance_sampling_ratio/mean": 0.03249359540641308,
+ "sampling/importance_sampling_ratio/min": 2.793159140067336e-05,
+ "sampling/sampling_logp_difference/max": 0.34277451038360596,
+ "sampling/sampling_logp_difference/mean": 0.01319133285433054,
+ "step": 410,
+ "step_time": 15.258313446864486
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000596046447,
+ "completions/max_length": 2048.0,
+ "completions/max_terminated_length": 1373.4,
+ "completions/mean_length": 1174.6666870117188,
+ "completions/mean_terminated_length": 982.5400207519531,
+ "completions/min_length": 717.8,
+ "completions/min_terminated_length": 717.8,
+ "entropy": 0.17395039709905782,
+ "epoch": 0.151183970856102,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.1571768969297409,
+ "kl": 0.0006294176423883376,
+ "learning_rate": 4.468941717791411e-07,
+ "loss": -0.013487279415130615,
+ "num_tokens": 2744805.0,
+ "reward": 1.6100000858306884,
+ "reward_std": 0.9736106634140015,
+ "rewards/boxed_rate/mean": 0.799999988079071,
+ "rewards/boxed_rate/std": 0.4298781991004944,
+ "rewards/correctness/mean": 0.3000000089406967,
+ "rewards/correctness/std": 0.3914883255958557,
+ "rewards/multi_component_reward/mean": 0.31000002175569535,
+ "rewards/multi_component_reward/std": 0.5971025705337525,
+ "rewards/no_answer_rate/mean": 0.20000000596046447,
+ "rewards/no_answer_rate/std": 0.4298781991004944,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.09501664619892836,
+ "sampling/importance_sampling_ratio/mean": 0.02351754056289792,
+ "sampling/importance_sampling_ratio/min": 1.3345446579433804e-05,
+ "sampling/sampling_logp_difference/max": 0.35430474281311036,
+ "sampling/sampling_logp_difference/mean": 0.012131126876920462,
+ "step": 415,
+ "step_time": 17.18821715861559
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1233.4,
+ "completions/max_terminated_length": 934.0,
+ "completions/mean_length": 838.5667053222656,
+ "completions/mean_terminated_length": 744.7300109863281,
+ "completions/min_length": 594.4,
+ "completions/min_terminated_length": 594.4,
+ "entropy": 0.17896714533368746,
+ "epoch": 0.15300546448087432,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.08533581346273422,
+ "kl": 0.00047911926230881363,
+ "learning_rate": 4.4593558282208586e-07,
+ "loss": -0.00437382236123085,
+ "num_tokens": 2771858.0,
+ "reward": 1.5300000667572022,
+ "reward_std": 0.8672845721244812,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.2333333373069763,
+ "rewards/correctness/std": 0.3761233925819397,
+ "rewards/multi_component_reward/mean": 0.29666668176651,
+ "rewards/multi_component_reward/std": 0.49711505472660067,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.07356768772006035,
+ "sampling/importance_sampling_ratio/mean": 0.026540174335241317,
+ "sampling/importance_sampling_ratio/min": 0.0027435134909879364,
+ "sampling/sampling_logp_difference/max": 0.2972338914871216,
+ "sampling/sampling_logp_difference/mean": 0.011641639284789562,
+ "step": 420,
+ "step_time": 11.845613337121904
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1578.6,
+ "completions/max_terminated_length": 1112.6,
+ "completions/mean_length": 957.2333740234375,
+ "completions/mean_terminated_length": 789.5533447265625,
+ "completions/min_length": 506.8,
+ "completions/min_terminated_length": 506.8,
+ "entropy": 0.22455186421672504,
+ "epoch": 0.15482695810564662,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.011058821342885494,
+ "kl": 0.0005187941482290625,
+ "learning_rate": 4.449769938650307e-07,
+ "loss": -0.0018246347084641457,
+ "num_tokens": 2803527.0,
+ "reward": 1.6033333539962769,
+ "reward_std": 0.8526834607124328,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.26666667461395266,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.3033333346247673,
+ "rewards/multi_component_reward/std": 0.5273210048675537,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.040888981963507834,
+ "sampling/importance_sampling_ratio/mean": 0.012350313179194928,
+ "sampling/importance_sampling_ratio/min": 4.1122279969854155e-05,
+ "sampling/sampling_logp_difference/max": 0.3840524673461914,
+ "sampling/sampling_logp_difference/mean": 0.014351606741547585,
+ "step": 425,
+ "step_time": 13.965443047881127
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1702.0,
+ "completions/max_terminated_length": 1172.4,
+ "completions/mean_length": 1089.1000244140625,
+ "completions/mean_terminated_length": 908.7400146484375,
+ "completions/min_length": 716.8,
+ "completions/min_terminated_length": 716.8,
+ "entropy": 0.1683452881872654,
+ "epoch": 0.15664845173041894,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.014243739657104015,
+ "kl": 0.0004299223937171822,
+ "learning_rate": 4.4401840490797544e-07,
+ "loss": -5.882064579054713e-05,
+ "num_tokens": 2838744.0,
+ "reward": 1.5133334517478942,
+ "reward_std": 0.6114970922470093,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.23333333134651185,
+ "rewards/correctness/std": 0.24494898319244385,
+ "rewards/multi_component_reward/mean": 0.24666669517755507,
+ "rewards/multi_component_reward/std": 0.39071274995803834,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.08568799644708633,
+ "sampling/importance_sampling_ratio/mean": 0.032788149639964105,
+ "sampling/importance_sampling_ratio/min": 0.0002142171015298561,
+ "sampling/sampling_logp_difference/max": 0.3399065971374512,
+ "sampling/sampling_logp_difference/mean": 0.011240929551422595,
+ "step": 430,
+ "step_time": 14.933206058479845
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.2666666716337204,
+ "completions/max_length": 1818.2,
+ "completions/max_terminated_length": 1232.2,
+ "completions/mean_length": 1174.3333618164063,
+ "completions/mean_terminated_length": 910.4933471679688,
+ "completions/min_length": 678.2,
+ "completions/min_terminated_length": 678.2,
+ "entropy": 0.1602409650882085,
+ "epoch": 0.15846994535519127,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.07029227167367935,
+ "kl": 0.00047036608011694623,
+ "learning_rate": 4.4305981595092023e-07,
+ "loss": -0.004705151170492172,
+ "num_tokens": 2877094.0,
+ "reward": 1.4300000429153443,
+ "reward_std": 0.7005062401294708,
+ "rewards/boxed_rate/mean": 0.7333333373069764,
+ "rewards/boxed_rate/std": 0.3977532863616943,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.13000000715255738,
+ "rewards/multi_component_reward/std": 0.5194472134113312,
+ "rewards/no_answer_rate/mean": 0.2666666716337204,
+ "rewards/no_answer_rate/std": 0.3977532863616943,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.24494898319244385,
+ "sampling/importance_sampling_ratio/max": 0.13022951409220695,
+ "sampling/importance_sampling_ratio/mean": 0.05298698227852583,
+ "sampling/importance_sampling_ratio/min": 9.12893192064404e-05,
+ "sampling/sampling_logp_difference/max": 0.3845913648605347,
+ "sampling/sampling_logp_difference/mean": 0.01090236771851778,
+ "step": 435,
+ "step_time": 15.83292675986886
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1450.6,
+ "completions/max_terminated_length": 1089.0,
+ "completions/mean_length": 925.2000366210938,
+ "completions/mean_terminated_length": 851.3733581542969,
+ "completions/min_length": 552.8,
+ "completions/min_terminated_length": 552.8,
+ "entropy": 0.19520388320088386,
+ "epoch": 0.16029143897996356,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.11756877601146698,
+ "kl": 0.0004913257944281213,
+ "learning_rate": 4.4210122699386497e-07,
+ "loss": -0.009752783924341202,
+ "num_tokens": 2907568.0,
+ "reward": 1.783333420753479,
+ "reward_std": 0.8172009825706482,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.33333333134651183,
+ "rewards/correctness/std": 0.35449349880218506,
+ "rewards/multi_component_reward/mean": 0.41666669249534605,
+ "rewards/multi_component_reward/std": 0.4753227561712265,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.07866446217522025,
+ "sampling/importance_sampling_ratio/mean": 0.03093269495293498,
+ "sampling/importance_sampling_ratio/min": 0.002900865976889124,
+ "sampling/sampling_logp_difference/max": 0.3607002258300781,
+ "sampling/sampling_logp_difference/mean": 0.01270847897976637,
+ "step": 440,
+ "step_time": 13.279750570654869
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1239.0,
+ "completions/max_terminated_length": 991.2,
+ "completions/mean_length": 793.8333618164063,
+ "completions/mean_terminated_length": 740.3466857910156,
+ "completions/min_length": 537.4,
+ "completions/min_terminated_length": 537.4,
+ "entropy": 0.172708131869634,
+ "epoch": 0.1621129326047359,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.1338849663734436,
+ "kl": 0.0005728138513707866,
+ "learning_rate": 4.411426380368098e-07,
+ "loss": -0.0031532272696495056,
+ "num_tokens": 2937617.0,
+ "reward": 1.6400001287460326,
+ "reward_std": 0.6522666096687317,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.2666666716337204,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.37333334386348727,
+ "rewards/multi_component_reward/std": 0.35980162024497986,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.07935073487460613,
+ "sampling/importance_sampling_ratio/mean": 0.024918343313038348,
+ "sampling/importance_sampling_ratio/min": 0.001050433111959137,
+ "sampling/sampling_logp_difference/max": 0.37157559394836426,
+ "sampling/sampling_logp_difference/mean": 0.011105079017579556,
+ "step": 445,
+ "step_time": 11.855768189579248
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1251.4,
+ "completions/max_terminated_length": 829.6,
+ "completions/mean_length": 769.7333618164063,
+ "completions/mean_terminated_length": 598.9666748046875,
+ "completions/min_length": 440.8,
+ "completions/min_terminated_length": 440.8,
+ "entropy": 0.16670421635111174,
+ "epoch": 0.16393442622950818,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.47747763991355896,
+ "kl": 0.00048284496394141265,
+ "learning_rate": 4.401840490797546e-07,
+ "loss": -0.013033266365528106,
+ "num_tokens": 2963841.0,
+ "reward": 1.9300000429153443,
+ "reward_std": 0.8960777290165425,
+ "rewards/boxed_rate/mean": 0.8666666746139526,
+ "rewards/boxed_rate/std": 0.20655910968780516,
+ "rewards/correctness/mean": 0.40000001192092893,
+ "rewards/correctness/std": 0.41311821937561033,
+ "rewards/multi_component_reward/mean": 0.42999999821186063,
+ "rewards/multi_component_reward/std": 0.601305240392685,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.20655910968780516,
+ "rewards/repetition_rate/mean": 0.10000000298023223,
+ "rewards/repetition_rate/std": 0.18492921590805053,
+ "sampling/importance_sampling_ratio/max": 0.14179300479590892,
+ "sampling/importance_sampling_ratio/mean": 0.06188145540654659,
+ "sampling/importance_sampling_ratio/min": 0.017718057095797223,
+ "sampling/sampling_logp_difference/max": 0.3239980101585388,
+ "sampling/sampling_logp_difference/mean": 0.010319609567523003,
+ "step": 450,
+ "step_time": 11.826861386746168
+ },
+ {
+ "epoch": 0.16393442622950818,
+ "eval_clip_ratio/high_max": 0.0,
+ "eval_clip_ratio/high_mean": 0.0,
+ "eval_clip_ratio/low_mean": 0.0,
+ "eval_clip_ratio/low_min": 0.0,
+ "eval_clip_ratio/region_mean": 0.0,
+ "eval_completions/clipped_ratio": 0.20666667073965073,
+ "eval_completions/max_length": 1688.02,
+ "eval_completions/max_terminated_length": 1185.42,
+ "eval_completions/mean_length": 1123.9933685302735,
+ "eval_completions/mean_terminated_length": 846.4790167236329,
+ "eval_completions/min_length": 676.62,
+ "eval_completions/min_terminated_length": 594.7,
+ "eval_entropy": 0.18630945876240732,
+ "eval_frac_reward_zero_std": 0.16,
+ "eval_kl": 0.0004587844273191877,
+ "eval_loss": -0.0019339508144184947,
+ "eval_num_tokens": 2963841.0,
+ "eval_reward": 1.5870000886917115,
+ "eval_reward_std": 0.7317568999528885,
+ "eval_rewards/boxed_rate/mean": 0.7933333313465118,
+ "eval_rewards/boxed_rate/std": 0.2689453566074371,
+ "eval_rewards/correctness/mean": 0.2733333346247673,
+ "eval_rewards/correctness/std": 0.2873582673072815,
+ "eval_rewards/multi_component_reward/mean": 0.2603333519771695,
+ "eval_rewards/multi_component_reward/std": 0.4510752035677433,
+ "eval_rewards/no_answer_rate/mean": 0.20666667073965073,
+ "eval_rewards/no_answer_rate/std": 0.2689453566074371,
+ "eval_rewards/repetition_rate/mean": 0.05333333492279053,
+ "eval_rewards/repetition_rate/std": 0.09630359530448913,
+ "eval_runtime": 669.3873,
+ "eval_samples_per_second": 0.075,
+ "eval_sampling/importance_sampling_ratio/max": 0.10398212056839838,
+ "eval_sampling/importance_sampling_ratio/mean": 0.035226596340653483,
+ "eval_sampling/importance_sampling_ratio/min": 0.0024896363094269146,
+ "eval_sampling/sampling_logp_difference/max": 0.5326641929149628,
+ "eval_sampling/sampling_logp_difference/mean": 0.012828467870131135,
+ "eval_steps_per_second": 0.013,
+ "step": 450
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1987.6,
+ "completions/max_terminated_length": 1503.0,
+ "completions/mean_length": 1150.6666870117188,
+ "completions/mean_terminated_length": 990.6999938964843,
+ "completions/min_length": 602.2,
+ "completions/min_terminated_length": 602.2,
+ "entropy": 0.2162185865143935,
+ "epoch": 0.1657559198542805,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.01271678227931261,
+ "kl": 0.0005497580714290962,
+ "learning_rate": 4.3922546012269933e-07,
+ "loss": 0.0011618516407907009,
+ "num_tokens": 3000941.0,
+ "reward": 1.6966667413711547,
+ "reward_std": 0.9000118732452392,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.3000000089406967,
+ "rewards/correctness/std": 0.3914883255958557,
+ "rewards/multi_component_reward/mean": 0.3300000101327896,
+ "rewards/multi_component_reward/std": 0.5810832887887954,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.11817909283563495,
+ "sampling/importance_sampling_ratio/mean": 0.02859312845394015,
+ "sampling/importance_sampling_ratio/min": 0.00019200471108481247,
+ "sampling/sampling_logp_difference/max": 0.40274404287338256,
+ "sampling/sampling_logp_difference/mean": 0.014035884011536836,
+ "step": 455,
+ "step_time": 16.644585143402217
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1850.4,
+ "completions/max_terminated_length": 1004.8,
+ "completions/mean_length": 914.5666870117187,
+ "completions/mean_terminated_length": 717.0500061035157,
+ "completions/min_length": 396.2,
+ "completions/min_terminated_length": 396.2,
+ "entropy": 0.18004637969036896,
+ "epoch": 0.16757741347905283,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.051200754940509796,
+ "kl": 0.0005335340790528183,
+ "learning_rate": 4.3826687116564417e-07,
+ "loss": -0.01714204400777817,
+ "num_tokens": 3030898.0,
+ "reward": 1.5366667270660401,
+ "reward_std": 0.7392153739929199,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.23666667342185974,
+ "rewards/multi_component_reward/std": 0.4870242774486542,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.1253473086282611,
+ "sampling/importance_sampling_ratio/mean": 0.06275567463599145,
+ "sampling/importance_sampling_ratio/min": 0.0048506295410769285,
+ "sampling/sampling_logp_difference/max": 0.35658023357391355,
+ "sampling/sampling_logp_difference/mean": 0.012955071125179528,
+ "step": 460,
+ "step_time": 15.270139714889229
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1409.8,
+ "completions/max_terminated_length": 850.4,
+ "completions/mean_length": 728.4666931152344,
+ "completions/mean_terminated_length": 629.5000183105469,
+ "completions/min_length": 469.8,
+ "completions/min_terminated_length": 469.8,
+ "entropy": 0.16985683316985767,
+ "epoch": 0.16939890710382513,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.02098464034497738,
+ "kl": 0.0005443188643160586,
+ "learning_rate": 4.3730828220858896e-07,
+ "loss": -0.007486963272094726,
+ "num_tokens": 3056190.0,
+ "reward": 1.9700001001358032,
+ "reward_std": 0.4863309383392334,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.43333333134651186,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.5366666868329049,
+ "rewards/multi_component_reward/std": 0.3014017343521118,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.12007005885243416,
+ "sampling/importance_sampling_ratio/mean": 0.05674390830099583,
+ "sampling/importance_sampling_ratio/min": 0.0004069998664532642,
+ "sampling/sampling_logp_difference/max": 0.5247022390365601,
+ "sampling/sampling_logp_difference/mean": 0.012214668095111847,
+ "step": 465,
+ "step_time": 12.395325169526041
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1901.0,
+ "completions/max_terminated_length": 1594.8,
+ "completions/mean_length": 1356.2000244140625,
+ "completions/mean_terminated_length": 1266.056689453125,
+ "completions/min_length": 993.0,
+ "completions/min_terminated_length": 993.0,
+ "entropy": 0.10780975048740706,
+ "epoch": 0.17122040072859745,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.05852176249027252,
+ "kl": 0.0004035008855377479,
+ "learning_rate": 4.363496932515337e-07,
+ "loss": -0.0037086054682731627,
+ "num_tokens": 3100044.0,
+ "reward": 1.6266667366027832,
+ "reward_std": 0.9011179387569428,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.3823883533477783,
+ "rewards/multi_component_reward/mean": 0.2933333463966846,
+ "rewards/multi_component_reward/std": 0.5466254830360413,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.062121391948312524,
+ "sampling/importance_sampling_ratio/mean": 0.0237305941991508,
+ "sampling/importance_sampling_ratio/min": 0.004560414611228225,
+ "sampling/sampling_logp_difference/max": 0.33158082962036134,
+ "sampling/sampling_logp_difference/mean": 0.007295731455087662,
+ "step": 470,
+ "step_time": 16.880542963929475
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1782.4,
+ "completions/max_terminated_length": 1526.4,
+ "completions/mean_length": 1103.5000427246093,
+ "completions/mean_terminated_length": 1016.5600524902344,
+ "completions/min_length": 741.8,
+ "completions/min_terminated_length": 741.8,
+ "entropy": 0.1433272872120142,
+ "epoch": 0.17304189435336975,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.03871724009513855,
+ "kl": 0.0003871701846946962,
+ "learning_rate": 4.353911042944785e-07,
+ "loss": -0.0024786345660686494,
+ "num_tokens": 3136959.0,
+ "reward": 1.1100000619888306,
+ "reward_std": 0.318433678150177,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.03333333432674408,
+ "rewards/correctness/std": 0.08164966106414795,
+ "rewards/multi_component_reward/mean": 0.0766666665673256,
+ "rewards/multi_component_reward/std": 0.23678402006626129,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.0617107892408967,
+ "sampling/importance_sampling_ratio/mean": 0.02746242703869939,
+ "sampling/importance_sampling_ratio/min": 8.897909594907105e-05,
+ "sampling/sampling_logp_difference/max": 0.3906729698181152,
+ "sampling/sampling_logp_difference/mean": 0.009637977462261915,
+ "step": 475,
+ "step_time": 15.594477642700076
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1867.2,
+ "completions/max_terminated_length": 1634.6,
+ "completions/mean_length": 1205.7000244140625,
+ "completions/mean_terminated_length": 989.493359375,
+ "completions/min_length": 550.8,
+ "completions/min_terminated_length": 550.8,
+ "entropy": 0.16859658484657605,
+ "epoch": 0.17486338797814208,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.0642157569527626,
+ "kl": 0.0005781898080992202,
+ "learning_rate": 4.3443251533742333e-07,
+ "loss": -0.0028929352760314942,
+ "num_tokens": 3175776.0,
+ "reward": 1.3066667556762694,
+ "reward_std": 0.6569535873830319,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.1066666692495346,
+ "rewards/multi_component_reward/std": 0.4782795637845993,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.06810643598437309,
+ "sampling/importance_sampling_ratio/mean": 0.01840525190345943,
+ "sampling/importance_sampling_ratio/min": 0.0006505777751514535,
+ "sampling/sampling_logp_difference/max": 0.43449815511703493,
+ "sampling/sampling_logp_difference/mean": 0.011115654464811086,
+ "step": 480,
+ "step_time": 15.972666030749679
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1663.0,
+ "completions/max_terminated_length": 1303.0,
+ "completions/mean_length": 1173.3333740234375,
+ "completions/mean_terminated_length": 989.41669921875,
+ "completions/min_length": 683.0,
+ "completions/min_terminated_length": 683.0,
+ "entropy": 0.14007456911106905,
+ "epoch": 0.1766848816029144,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.010275622829794884,
+ "kl": 0.00036965057806810365,
+ "learning_rate": 4.3347392638036806e-07,
+ "loss": 0.0001720041735097766,
+ "num_tokens": 3214522.0,
+ "reward": 1.9766667842864991,
+ "reward_std": 0.9272954225540161,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.43333333134651186,
+ "rewards/correctness/std": 0.404018247127533,
+ "rewards/multi_component_reward/mean": 0.47666670083999635,
+ "rewards/multi_component_reward/std": 0.5632504820823669,
+ "rewards/no_answer_rate/mean": 0.13333333432674407,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.05611779941245913,
+ "sampling/importance_sampling_ratio/mean": 0.021316760312765838,
+ "sampling/importance_sampling_ratio/min": 0.0011236143288555823,
+ "sampling/sampling_logp_difference/max": 0.3524420738220215,
+ "sampling/sampling_logp_difference/mean": 0.00864975331351161,
+ "step": 485,
+ "step_time": 15.02983694653958
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1206.8,
+ "completions/max_terminated_length": 934.0,
+ "completions/mean_length": 728.1666748046875,
+ "completions/mean_terminated_length": 595.9966796875,
+ "completions/min_length": 426.2,
+ "completions/min_terminated_length": 426.2,
+ "entropy": 0.23062201142311095,
+ "epoch": 0.1785063752276867,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.018157748505473137,
+ "kl": 0.001033562931115739,
+ "learning_rate": 4.3251533742331285e-07,
+ "loss": 0.014268814027309418,
+ "num_tokens": 3238755.0,
+ "reward": 1.833333384990692,
+ "reward_std": 0.6647640764713287,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.3666666686534882,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.43333335071802137,
+ "rewards/multi_component_reward/std": 0.4189693987369537,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.12710050344467164,
+ "sampling/importance_sampling_ratio/mean": 0.05628317659720779,
+ "sampling/importance_sampling_ratio/min": 0.015683322838595436,
+ "sampling/sampling_logp_difference/max": 0.36403354406356814,
+ "sampling/sampling_logp_difference/mean": 0.015313766896724701,
+ "step": 490,
+ "step_time": 11.357261310145258
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1498.2,
+ "completions/max_terminated_length": 1076.4,
+ "completions/mean_length": 1013.5000305175781,
+ "completions/mean_terminated_length": 849.3533508300782,
+ "completions/min_length": 602.6,
+ "completions/min_terminated_length": 602.6,
+ "entropy": 0.14790264442563056,
+ "epoch": 0.18032786885245902,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.034156352281570435,
+ "kl": 0.0005029304617589029,
+ "learning_rate": 4.315567484662577e-07,
+ "loss": -0.010992055386304855,
+ "num_tokens": 3274110.0,
+ "reward": 1.4433334589004516,
+ "reward_std": 0.6818739175796509,
+ "rewards/boxed_rate/mean": 0.8333333313465119,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.21000000536441804,
+ "rewards/multi_component_reward/std": 0.4088459849357605,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.10836826749145985,
+ "sampling/importance_sampling_ratio/mean": 0.04134368039667606,
+ "sampling/importance_sampling_ratio/min": 0.0007355699201900378,
+ "sampling/sampling_logp_difference/max": 0.5285765409469605,
+ "sampling/sampling_logp_difference/mean": 0.00972840515896678,
+ "step": 495,
+ "step_time": 13.717384218610823
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1768.6,
+ "completions/max_terminated_length": 1172.4,
+ "completions/mean_length": 1021.0333618164062,
+ "completions/mean_terminated_length": 829.9766723632813,
+ "completions/min_length": 574.6,
+ "completions/min_terminated_length": 574.6,
+ "entropy": 0.1277496966222922,
+ "epoch": 0.18214936247723132,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.036998555064201355,
+ "kl": 0.00041782591021425713,
+ "learning_rate": 4.3059815950920243e-07,
+ "loss": -0.014814507961273194,
+ "num_tokens": 3306649.0,
+ "reward": 1.9100001335144043,
+ "reward_std": 0.8856898427009583,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.4333333373069763,
+ "rewards/correctness/std": 0.3761233925819397,
+ "rewards/multi_component_reward/mean": 0.4766666904091835,
+ "rewards/multi_component_reward/std": 0.5212879002094268,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.2542514935135841,
+ "sampling/importance_sampling_ratio/mean": 0.09570314064621925,
+ "sampling/importance_sampling_ratio/min": 0.0020768470163739948,
+ "sampling/sampling_logp_difference/max": 0.34144341945648193,
+ "sampling/sampling_logp_difference/mean": 0.010077364556491376,
+ "step": 500,
+ "step_time": 15.174134342558682
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.0,
+ "completions/max_length": 1232.6,
+ "completions/max_terminated_length": 1232.6,
+ "completions/mean_length": 928.7333618164063,
+ "completions/mean_terminated_length": 928.7333618164063,
+ "completions/min_length": 612.8,
+ "completions/min_terminated_length": 612.8,
+ "entropy": 0.15919652469456197,
+ "epoch": 0.18397085610200364,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.03940599039196968,
+ "kl": 0.0004976143633636335,
+ "learning_rate": 4.296395705521472e-07,
+ "loss": 8.051458280533552e-05,
+ "num_tokens": 3338765.0,
+ "reward": 1.6133334398269654,
+ "reward_std": 0.5944257378578186,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.34666666090488435,
+ "rewards/multi_component_reward/std": 0.3269805133342743,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.12865251041948794,
+ "sampling/importance_sampling_ratio/mean": 0.046758548077195886,
+ "sampling/importance_sampling_ratio/min": 0.001370607664284762,
+ "sampling/sampling_logp_difference/max": 0.37642626762390136,
+ "sampling/sampling_logp_difference/mean": 0.010083456197753549,
+ "step": 505,
+ "step_time": 12.115066050365566
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1233.6,
+ "completions/max_terminated_length": 1011.6,
+ "completions/mean_length": 741.0666870117187,
+ "completions/mean_terminated_length": 699.6933471679688,
+ "completions/min_length": 502.8,
+ "completions/min_terminated_length": 502.8,
+ "entropy": 0.14653264259298643,
+ "epoch": 0.18579234972677597,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.022397559136152267,
+ "kl": 0.0004972440418593275,
+ "learning_rate": 4.28680981595092e-07,
+ "loss": -0.010462169349193574,
+ "num_tokens": 3363841.0,
+ "reward": 2.0600001335144045,
+ "reward_std": 0.9187898755073547,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.4666666626930237,
+ "rewards/correctness/std": 0.4298781991004944,
+ "rewards/multi_component_reward/mean": 0.5933333843946457,
+ "rewards/multi_component_reward/std": 0.49362565875053405,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.16412589382380247,
+ "sampling/importance_sampling_ratio/mean": 0.0629353643860668,
+ "sampling/importance_sampling_ratio/min": 0.011790907313888965,
+ "sampling/sampling_logp_difference/max": 0.3437652587890625,
+ "sampling/sampling_logp_difference/mean": 0.009886026009917259,
+ "step": 510,
+ "step_time": 11.563566098548472
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1427.2,
+ "completions/max_terminated_length": 666.4,
+ "completions/mean_length": 705.2000061035156,
+ "completions/mean_terminated_length": 511.7700134277344,
+ "completions/min_length": 344.2,
+ "completions/min_terminated_length": 344.2,
+ "entropy": 0.26261382003625233,
+ "epoch": 0.18761384335154827,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.10859289020299911,
+ "kl": 0.0008684398761639992,
+ "learning_rate": 4.277223926380368e-07,
+ "loss": -0.002386706694960594,
+ "num_tokens": 3386899.0,
+ "reward": 1.3700001001358033,
+ "reward_std": 0.7070007205009461,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.1666666716337204,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.20333334654569626,
+ "rewards/multi_component_reward/std": 0.4274008393287659,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.07567991334944964,
+ "sampling/importance_sampling_ratio/mean": 0.024421159457415343,
+ "sampling/importance_sampling_ratio/min": 0.001837286539412988,
+ "sampling/sampling_logp_difference/max": 0.34601569175720215,
+ "sampling/sampling_logp_difference/mean": 0.017737336456775665,
+ "step": 515,
+ "step_time": 12.585359536856412
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1618.2,
+ "completions/max_terminated_length": 1045.2,
+ "completions/mean_length": 936.9666870117187,
+ "completions/mean_terminated_length": 758.3166870117187,
+ "completions/min_length": 469.8,
+ "completions/min_terminated_length": 469.8,
+ "entropy": 0.1936116027335326,
+ "epoch": 0.1894353369763206,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.0316350944340229,
+ "kl": 0.0006832293069843824,
+ "learning_rate": 4.267638036809816e-07,
+ "loss": -0.008644417673349381,
+ "num_tokens": 3417600.0,
+ "reward": 1.5800000786781312,
+ "reward_std": 0.7550430953502655,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.2666666716337204,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.3133333310484886,
+ "rewards/multi_component_reward/std": 0.46337632834911346,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.0883142702281475,
+ "sampling/importance_sampling_ratio/mean": 0.030587387550622226,
+ "sampling/importance_sampling_ratio/min": 0.0063572791577344466,
+ "sampling/sampling_logp_difference/max": 0.3835047721862793,
+ "sampling/sampling_logp_difference/mean": 0.013410749472677707,
+ "step": 520,
+ "step_time": 13.994114908203482
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1382.6,
+ "completions/max_terminated_length": 1288.6,
+ "completions/mean_length": 903.3000122070313,
+ "completions/mean_terminated_length": 852.4833374023438,
+ "completions/min_length": 593.0,
+ "completions/min_terminated_length": 593.0,
+ "entropy": 0.18077843909462293,
+ "epoch": 0.1912568306010929,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.024261396378278732,
+ "kl": 0.0005541147484715718,
+ "learning_rate": 4.2580521472392637e-07,
+ "loss": -0.0031598765403032304,
+ "num_tokens": 3448035.0,
+ "reward": 1.8300000786781312,
+ "reward_std": 0.6672061026096344,
+ "rewards/boxed_rate/mean": 0.9333333373069763,
+ "rewards/boxed_rate/std": 0.10327955484390258,
+ "rewards/correctness/mean": 0.3666666686534882,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.4633333504199982,
+ "rewards/multi_component_reward/std": 0.378997391462326,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.10327955484390258,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.10047482601366937,
+ "sampling/importance_sampling_ratio/mean": 0.02871947007952258,
+ "sampling/importance_sampling_ratio/min": 0.0007279034301063802,
+ "sampling/sampling_logp_difference/max": 0.41403778791427615,
+ "sampling/sampling_logp_difference/mean": 0.011369843780994416,
+ "step": 525,
+ "step_time": 12.759401306509972
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000596046447,
+ "completions/max_length": 1729.4,
+ "completions/max_terminated_length": 1090.4,
+ "completions/mean_length": 1117.200048828125,
+ "completions/mean_terminated_length": 875.8833618164062,
+ "completions/min_length": 682.2,
+ "completions/min_terminated_length": 682.2,
+ "entropy": 0.18201239357391993,
+ "epoch": 0.1930783242258652,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.00892422441393137,
+ "kl": 0.000488846098111632,
+ "learning_rate": 4.2484662576687116e-07,
+ "loss": -0.004933690279722213,
+ "num_tokens": 3483861.0,
+ "reward": 1.3033334016799927,
+ "reward_std": 0.6930439114570618,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.34822853803634646,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.13666666969656943,
+ "rewards/multi_component_reward/std": 0.4622148424386978,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.34822853803634646,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.06577782724052668,
+ "sampling/importance_sampling_ratio/mean": 0.02595797423273325,
+ "sampling/importance_sampling_ratio/min": 0.0002929158277353877,
+ "sampling/sampling_logp_difference/max": 0.355899715423584,
+ "sampling/sampling_logp_difference/mean": 0.012041485961526632,
+ "step": 530,
+ "step_time": 15.400034189224243
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1707.4,
+ "completions/max_terminated_length": 958.8,
+ "completions/mean_length": 998.1000122070312,
+ "completions/mean_terminated_length": 766.7933349609375,
+ "completions/min_length": 525.4,
+ "completions/min_terminated_length": 525.4,
+ "entropy": 0.22935964514811832,
+ "epoch": 0.19489981785063754,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.020105650648474693,
+ "kl": 0.000557902626072367,
+ "learning_rate": 4.2388803680981595e-07,
+ "loss": -0.0007442331407219172,
+ "num_tokens": 3518562.0,
+ "reward": 1.2100000977516174,
+ "reward_std": 0.6208420097827911,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.10000000298023223,
+ "rewards/correctness/std": 0.24494898319244385,
+ "rewards/multi_component_reward/mean": 0.11000000834465026,
+ "rewards/multi_component_reward/std": 0.38656076192855837,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.015947306994348764,
+ "sampling/importance_sampling_ratio/mean": 0.004562706989236176,
+ "sampling/importance_sampling_ratio/min": 2.5106548244674746e-05,
+ "sampling/sampling_logp_difference/max": 0.36506946086883546,
+ "sampling/sampling_logp_difference/mean": 0.014524615183472634,
+ "step": 535,
+ "step_time": 14.785470444336534
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 2011.6,
+ "completions/max_terminated_length": 1252.0,
+ "completions/mean_length": 974.2000366210938,
+ "completions/mean_terminated_length": 799.5733703613281,
+ "completions/min_length": 499.0,
+ "completions/min_terminated_length": 499.0,
+ "entropy": 0.20504479110240936,
+ "epoch": 0.19672131147540983,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.016418447718024254,
+ "kl": 0.0007665569457458333,
+ "learning_rate": 4.229294478527607e-07,
+ "loss": -0.004444469138979912,
+ "num_tokens": 3550332.0,
+ "reward": 1.5333333969116212,
+ "reward_std": 0.6269734501838684,
+ "rewards/boxed_rate/mean": 0.8666666507720947,
+ "rewards/boxed_rate/std": 0.3265986442565918,
+ "rewards/correctness/mean": 0.23333333134651185,
+ "rewards/correctness/std": 0.24494898319244385,
+ "rewards/multi_component_reward/mean": 0.266666679084301,
+ "rewards/multi_component_reward/std": 0.4252805233001709,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.3265986442565918,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.0710077840834856,
+ "sampling/importance_sampling_ratio/mean": 0.0242210254073143,
+ "sampling/importance_sampling_ratio/min": 2.3673296637852498e-05,
+ "sampling/sampling_logp_difference/max": 0.37494783401489257,
+ "sampling/sampling_logp_difference/mean": 0.013403672352433204,
+ "step": 540,
+ "step_time": 16.360854005627335
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1755.4,
+ "completions/max_terminated_length": 1206.4,
+ "completions/mean_length": 827.3000122070313,
+ "completions/mean_terminated_length": 693.3333374023438,
+ "completions/min_length": 476.2,
+ "completions/min_terminated_length": 476.2,
+ "entropy": 0.20222537256777287,
+ "epoch": 0.19854280510018216,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.07017825543880463,
+ "kl": 0.0005556545128153327,
+ "learning_rate": 4.219708588957055e-07,
+ "loss": -0.0067356817424297334,
+ "num_tokens": 3577779.0,
+ "reward": 1.8100000619888306,
+ "reward_std": 0.7233119249343872,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.3666666686534882,
+ "rewards/correctness/std": 0.2882087707519531,
+ "rewards/multi_component_reward/mean": 0.44333335757255554,
+ "rewards/multi_component_reward/std": 0.43861203789711,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.07237468343228101,
+ "sampling/importance_sampling_ratio/mean": 0.02907364722341299,
+ "sampling/importance_sampling_ratio/min": 0.003791734886874308,
+ "sampling/sampling_logp_difference/max": 0.3440469026565552,
+ "sampling/sampling_logp_difference/mean": 0.013698535226285457,
+ "step": 545,
+ "step_time": 14.568762712925672
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1740.6,
+ "completions/max_terminated_length": 1098.0,
+ "completions/mean_length": 972.5000366210937,
+ "completions/mean_terminated_length": 813.2100280761719,
+ "completions/min_length": 548.8,
+ "completions/min_terminated_length": 548.8,
+ "entropy": 0.15388319492340088,
+ "epoch": 0.20036429872495445,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.09412358701229095,
+ "kl": 0.00047789589831760774,
+ "learning_rate": 4.210122699386503e-07,
+ "loss": 0.0004236143082380295,
+ "num_tokens": 3610650.0,
+ "reward": 1.7666667222976684,
+ "reward_std": 0.748459929227829,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.3333333432674408,
+ "rewards/correctness/std": 0.3098386645317078,
+ "rewards/multi_component_reward/mean": 0.3666666656732559,
+ "rewards/multi_component_reward/std": 0.5095339387655258,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.07994540650397539,
+ "sampling/importance_sampling_ratio/mean": 0.0390994424931705,
+ "sampling/importance_sampling_ratio/min": 0.0009056368981219976,
+ "sampling/sampling_logp_difference/max": 0.3753530740737915,
+ "sampling/sampling_logp_difference/mean": 0.009989727940410375,
+ "step": 550,
+ "step_time": 15.042225486785174
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1604.4,
+ "completions/max_terminated_length": 979.4,
+ "completions/mean_length": 832.6666870117188,
+ "completions/mean_terminated_length": 665.1333374023437,
+ "completions/min_length": 452.0,
+ "completions/min_terminated_length": 452.0,
+ "entropy": 0.2048481193681558,
+ "epoch": 0.20218579234972678,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.08716036379337311,
+ "kl": 0.000548629568462881,
+ "learning_rate": 4.2005368098159505e-07,
+ "loss": -0.005523241683840751,
+ "num_tokens": 3638024.0,
+ "reward": 1.6033334016799927,
+ "reward_std": 0.797308748960495,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.32236858606338503,
+ "rewards/multi_component_reward/mean": 0.30333334803581236,
+ "rewards/multi_component_reward/std": 0.5069476455450058,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.07022089585661888,
+ "sampling/importance_sampling_ratio/mean": 0.022322331066243352,
+ "sampling/importance_sampling_ratio/min": 0.000656013962759859,
+ "sampling/sampling_logp_difference/max": 0.3337650179862976,
+ "sampling/sampling_logp_difference/mean": 0.013166856206953525,
+ "step": 555,
+ "step_time": 13.850998834148049
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1198.2,
+ "completions/max_terminated_length": 1077.2,
+ "completions/mean_length": 757.1333618164062,
+ "completions/mean_terminated_length": 690.2466918945313,
+ "completions/min_length": 448.2,
+ "completions/min_terminated_length": 448.2,
+ "entropy": 0.18011298303802808,
+ "epoch": 0.2040072859744991,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.01096227765083313,
+ "kl": 0.0006569843031077956,
+ "learning_rate": 4.1909509202453984e-07,
+ "loss": -0.007073364406824112,
+ "num_tokens": 3663726.0,
+ "reward": 1.9000000715255738,
+ "reward_std": 0.8417381644248962,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.40000000298023225,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.5000000268220901,
+ "rewards/multi_component_reward/std": 0.4765937089920044,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.09752159416675568,
+ "sampling/importance_sampling_ratio/mean": 0.05281178932636976,
+ "sampling/importance_sampling_ratio/min": 0.003194372950383695,
+ "sampling/sampling_logp_difference/max": 0.3607391357421875,
+ "sampling/sampling_logp_difference/mean": 0.011687950044870377,
+ "step": 560,
+ "step_time": 11.587789290212095
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1599.4,
+ "completions/max_terminated_length": 1067.8,
+ "completions/mean_length": 1085.8000244140626,
+ "completions/mean_terminated_length": 867.1133544921875,
+ "completions/min_length": 651.6,
+ "completions/min_terminated_length": 651.6,
+ "entropy": 0.12313016926248868,
+ "epoch": 0.2058287795992714,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.05608590692281723,
+ "kl": 0.00031730876992999886,
+ "learning_rate": 4.181365030674847e-07,
+ "loss": 0.0087483711540699,
+ "num_tokens": 3699090.0,
+ "reward": 1.7966667652130126,
+ "reward_std": 0.8426974892616272,
+ "rewards/boxed_rate/mean": 0.8,
+ "rewards/boxed_rate/std": 0.29447373151779177,
+ "rewards/correctness/mean": 0.36666667461395264,
+ "rewards/correctness/std": 0.3161036252975464,
+ "rewards/multi_component_reward/mean": 0.36333334594964983,
+ "rewards/multi_component_reward/std": 0.5058179676532746,
+ "rewards/no_answer_rate/mean": 0.20000000298023224,
+ "rewards/no_answer_rate/std": 0.29447373151779177,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.10868240725831128,
+ "sampling/importance_sampling_ratio/mean": 0.04146801241877256,
+ "sampling/importance_sampling_ratio/min": 0.008836518857719966,
+ "sampling/sampling_logp_difference/max": 0.3564323544502258,
+ "sampling/sampling_logp_difference/mean": 0.007883247546851635,
+ "step": 565,
+ "step_time": 14.454759103059768
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.33333333432674406,
+ "completions/max_length": 1746.6,
+ "completions/max_terminated_length": 973.4,
+ "completions/mean_length": 1263.9000244140625,
+ "completions/mean_terminated_length": 730.0800231933594,
+ "completions/min_length": 952.8,
+ "completions/min_terminated_length": 543.2,
+ "entropy": 0.14238746762275695,
+ "epoch": 0.20765027322404372,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.01621309295296669,
+ "kl": 0.0003668556613168524,
+ "learning_rate": 4.171779141104294e-07,
+ "loss": 0.00620177686214447,
+ "num_tokens": 3740091.0,
+ "reward": 1.4800001382827759,
+ "reward_std": 0.5089709401130676,
+ "rewards/boxed_rate/mean": 0.6666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.2666666626930237,
+ "rewards/correctness/std": 0.19119417667388916,
+ "rewards/multi_component_reward/mean": 0.17999999970197678,
+ "rewards/multi_component_reward/std": 0.30394219756126406,
+ "rewards/no_answer_rate/mean": 0.33333333432674406,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.09319847710430622,
+ "sampling/importance_sampling_ratio/mean": 0.04575341045856476,
+ "sampling/importance_sampling_ratio/min": 0.004630656010958226,
+ "sampling/sampling_logp_difference/max": 0.3321978569030762,
+ "sampling/sampling_logp_difference/mean": 0.009725131979212164,
+ "step": 570,
+ "step_time": 15.637936590984463
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333432674407,
+ "completions/max_length": 1493.2,
+ "completions/max_terminated_length": 1230.4,
+ "completions/mean_length": 1086.3333618164063,
+ "completions/mean_terminated_length": 990.8466918945312,
+ "completions/min_length": 756.6,
+ "completions/min_terminated_length": 756.6,
+ "entropy": 0.1577371135354042,
+ "epoch": 0.20947176684881602,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.017710594460368156,
+ "kl": 0.0003999709859878446,
+ "learning_rate": 4.162193251533742e-07,
+ "loss": -0.0008659601211547851,
+ "num_tokens": 3778573.0,
+ "reward": 1.2533334612846374,
+ "reward_std": 0.6350648760795593,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.19119417667388916,
+ "rewards/correctness/mean": 0.10000000298023223,
+ "rewards/correctness/std": 0.24494898319244385,
+ "rewards/multi_component_reward/mean": 0.12000000327825547,
+ "rewards/multi_component_reward/std": 0.34919504523277284,
+ "rewards/no_answer_rate/mean": 0.13333333432674407,
+ "rewards/no_answer_rate/std": 0.19119417667388916,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.021646471600979565,
+ "sampling/importance_sampling_ratio/mean": 0.009971876302734018,
+ "sampling/importance_sampling_ratio/min": 0.0019394792594557403,
+ "sampling/sampling_logp_difference/max": 0.331281840801239,
+ "sampling/sampling_logp_difference/mean": 0.01047133533284068,
+ "step": 575,
+ "step_time": 13.984978386759758
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1077.0,
+ "completions/max_terminated_length": 843.4,
+ "completions/mean_length": 618.633349609375,
+ "completions/mean_terminated_length": 570.0800170898438,
+ "completions/min_length": 312.4,
+ "completions/min_terminated_length": 312.4,
+ "entropy": 0.18354702678819498,
+ "epoch": 0.21129326047358835,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.01734095625579357,
+ "kl": 0.0007040846471985181,
+ "learning_rate": 4.15260736196319e-07,
+ "loss": -0.034613341093063354,
+ "num_tokens": 3799994.0,
+ "reward": 1.733333420753479,
+ "reward_std": 0.8469935655593872,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.3000000029802322,
+ "rewards/correctness/std": 0.404018247127533,
+ "rewards/multi_component_reward/mean": 0.4000000163912773,
+ "rewards/multi_component_reward/std": 0.481517493724823,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.2370181828737259,
+ "sampling/importance_sampling_ratio/mean": 0.07491586618125438,
+ "sampling/importance_sampling_ratio/min": 0.017679853770823685,
+ "sampling/sampling_logp_difference/max": 0.2730631709098816,
+ "sampling/sampling_logp_difference/mean": 0.011196557991206646,
+ "step": 580,
+ "step_time": 10.380646949820221
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1423.0,
+ "completions/max_terminated_length": 1241.8,
+ "completions/mean_length": 999.9000366210937,
+ "completions/mean_terminated_length": 926.8766967773438,
+ "completions/min_length": 742.8,
+ "completions/min_terminated_length": 742.8,
+ "entropy": 0.17347905735174815,
+ "epoch": 0.21311475409836064,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.019545914605259895,
+ "kl": 0.0004457860350764046,
+ "learning_rate": 4.143021472392638e-07,
+ "loss": -0.0001349434838630259,
+ "num_tokens": 3834629.0,
+ "reward": 1.4100001215934754,
+ "reward_std": 0.6349397718906402,
+ "rewards/boxed_rate/mean": 0.9333333373069763,
+ "rewards/boxed_rate/std": 0.10327955484390258,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.2433333396911621,
+ "rewards/multi_component_reward/std": 0.36209596991539,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.10327955484390258,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.04343400467187166,
+ "sampling/importance_sampling_ratio/mean": 0.013024529488757253,
+ "sampling/importance_sampling_ratio/min": 0.001408253621775657,
+ "sampling/sampling_logp_difference/max": 0.5604913473129273,
+ "sampling/sampling_logp_difference/mean": 0.010900301579385995,
+ "step": 585,
+ "step_time": 13.423828819952906
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1522.6,
+ "completions/max_terminated_length": 1304.6,
+ "completions/mean_length": 1048.300048828125,
+ "completions/mean_terminated_length": 953.2133544921875,
+ "completions/min_length": 677.2,
+ "completions/min_terminated_length": 677.2,
+ "entropy": 0.13777906013031802,
+ "epoch": 0.21493624772313297,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.020055213943123817,
+ "kl": 0.00046802494471194224,
+ "learning_rate": 4.1334355828220857e-07,
+ "loss": 0.0007817570120096206,
+ "num_tokens": 3868946.0,
+ "reward": 1.3900001525878907,
+ "reward_std": 0.7840524554252625,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.1666666716337204,
+ "rewards/correctness/std": 0.34822853803634646,
+ "rewards/multi_component_reward/mean": 0.22333333939313887,
+ "rewards/multi_component_reward/std": 0.44443279504776,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.05379525385797024,
+ "sampling/importance_sampling_ratio/mean": 0.019492477364838123,
+ "sampling/importance_sampling_ratio/min": 0.0019421720592784908,
+ "sampling/sampling_logp_difference/max": 0.36746039390563967,
+ "sampling/sampling_logp_difference/mean": 0.009275554586201907,
+ "step": 590,
+ "step_time": 13.933117091469466
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1637.0,
+ "completions/max_terminated_length": 1173.6,
+ "completions/mean_length": 936.9000122070313,
+ "completions/mean_terminated_length": 814.3366760253906,
+ "completions/min_length": 551.2,
+ "completions/min_terminated_length": 551.2,
+ "entropy": 0.1666003334025542,
+ "epoch": 0.2167577413479053,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.011394626460969448,
+ "kl": 0.00048624937092730153,
+ "learning_rate": 4.1238496932515336e-07,
+ "loss": -6.75798102747649e-05,
+ "num_tokens": 3899387.0,
+ "reward": 1.483333444595337,
+ "reward_std": 0.6897397696971893,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.2500000074505806,
+ "rewards/multi_component_reward/std": 0.42727351784706114,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.03627656940370798,
+ "sampling/importance_sampling_ratio/mean": 0.014781518559902906,
+ "sampling/importance_sampling_ratio/min": 0.0008682889689286911,
+ "sampling/sampling_logp_difference/max": 0.3049093961715698,
+ "sampling/sampling_logp_difference/mean": 0.010832818690687418,
+ "step": 595,
+ "step_time": 14.285567329451442
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1661.6,
+ "completions/max_terminated_length": 1382.2,
+ "completions/mean_length": 1118.2000244140625,
+ "completions/mean_terminated_length": 938.136669921875,
+ "completions/min_length": 657.6,
+ "completions/min_terminated_length": 657.6,
+ "entropy": 0.17573260565598806,
+ "epoch": 0.2185792349726776,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.020734934136271477,
+ "kl": 0.0004598392901243642,
+ "learning_rate": 4.1142638036809815e-07,
+ "loss": -0.002467634528875351,
+ "num_tokens": 3936269.0,
+ "reward": 1.0933334231376648,
+ "reward_std": 0.33367283940315245,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.03333333432674408,
+ "rewards/correctness/std": 0.08164966106414795,
+ "rewards/multi_component_reward/mean": 0.02666666656732559,
+ "rewards/multi_component_reward/std": 0.2805217713117599,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.044962253980338575,
+ "sampling/importance_sampling_ratio/mean": 0.011017974140122533,
+ "sampling/importance_sampling_ratio/min": 0.0003941454909745216,
+ "sampling/sampling_logp_difference/max": 0.38812339305877686,
+ "sampling/sampling_logp_difference/mean": 0.01122636515647173,
+ "step": 600,
+ "step_time": 14.878706483915447
+ },
+ {
+ "epoch": 0.2185792349726776,
+ "eval_clip_ratio/high_max": 0.0,
+ "eval_clip_ratio/high_mean": 0.0,
+ "eval_clip_ratio/low_mean": 0.0,
+ "eval_clip_ratio/low_min": 0.0,
+ "eval_clip_ratio/region_mean": 0.0,
+ "eval_completions/clipped_ratio": 0.18333333790302275,
+ "eval_completions/max_length": 1633.0,
+ "eval_completions/max_terminated_length": 1261.48,
+ "eval_completions/mean_length": 1074.5033642578126,
+ "eval_completions/mean_terminated_length": 884.9103503417969,
+ "eval_completions/min_length": 608.76,
+ "eval_completions/min_terminated_length": 608.76,
+ "eval_entropy": 0.1780784809589386,
+ "eval_frac_reward_zero_std": 0.1,
+ "eval_kl": 0.000503081139177084,
+ "eval_loss": -0.002404627623036504,
+ "eval_num_tokens": 3936269.0,
+ "eval_reward": 1.5616667437553406,
+ "eval_reward_std": 0.7698906905949116,
+ "eval_rewards/boxed_rate/mean": 0.816666665673256,
+ "eval_rewards/boxed_rate/std": 0.28143630146980286,
+ "eval_rewards/correctness/mean": 0.260000002682209,
+ "eval_rewards/correctness/std": 0.3175121474266052,
+ "eval_rewards/multi_component_reward/mean": 0.26500001683831215,
+ "eval_rewards/multi_component_reward/std": 0.49374800592660906,
+ "eval_rewards/no_answer_rate/mean": 0.18333333790302275,
+ "eval_rewards/no_answer_rate/std": 0.28143630146980286,
+ "eval_rewards/repetition_rate/mean": 0.036666667759418486,
+ "eval_rewards/repetition_rate/std": 0.08381265044212341,
+ "eval_runtime": 648.1074,
+ "eval_samples_per_second": 0.077,
+ "eval_sampling/importance_sampling_ratio/max": 0.08480258989613504,
+ "eval_sampling/importance_sampling_ratio/mean": 0.03125391862238757,
+ "eval_sampling/importance_sampling_ratio/min": 0.0025257051857090485,
+ "eval_sampling/sampling_logp_difference/max": 0.5640374529361725,
+ "eval_sampling/sampling_logp_difference/mean": 0.01284005825407803,
+ "eval_steps_per_second": 0.014,
+ "step": 600
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1519.2,
+ "completions/max_terminated_length": 1256.6,
+ "completions/mean_length": 946.06669921875,
+ "completions/mean_terminated_length": 839.1633544921875,
+ "completions/min_length": 563.2,
+ "completions/min_terminated_length": 563.2,
+ "entropy": 0.14678182030717532,
+ "epoch": 0.2204007285974499,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.023046690970659256,
+ "kl": 0.0005303417919397664,
+ "learning_rate": 4.1046779141104294e-07,
+ "loss": -0.0009497999213635922,
+ "num_tokens": 3967489.0,
+ "reward": 1.2500001072883606,
+ "reward_std": 0.46636478304862977,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.10000000298023223,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.15000001192092896,
+ "rewards/multi_component_reward/std": 0.28614950776100156,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.15675322916358708,
+ "sampling/importance_sampling_ratio/mean": 0.057061844225972894,
+ "sampling/importance_sampling_ratio/min": 0.0010045379561626475,
+ "sampling/sampling_logp_difference/max": 0.3354309320449829,
+ "sampling/sampling_logp_difference/mean": 0.010376049391925334,
+ "step": 605,
+ "step_time": 13.543810763396323
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1038.6,
+ "completions/max_terminated_length": 971.6,
+ "completions/mean_length": 741.6333618164062,
+ "completions/mean_terminated_length": 710.4466918945312,
+ "completions/min_length": 523.2,
+ "completions/min_terminated_length": 523.2,
+ "entropy": 0.17811203648646673,
+ "epoch": 0.2222222222222222,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 1.1156023740768433,
+ "kl": 0.0007145934030025577,
+ "learning_rate": 4.095092024539877e-07,
+ "loss": 0.001612265408039093,
+ "num_tokens": 3992006.0,
+ "reward": 1.8733334302902223,
+ "reward_std": 0.8630595266819,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.3666666656732559,
+ "rewards/correctness/std": 0.3761233925819397,
+ "rewards/multi_component_reward/mean": 0.4733333714306355,
+ "rewards/multi_component_reward/std": 0.4589902624487877,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.12649268209934234,
+ "sampling/importance_sampling_ratio/mean": 0.06289250953122974,
+ "sampling/importance_sampling_ratio/min": 0.03086609955687436,
+ "sampling/sampling_logp_difference/max": 0.3575249195098877,
+ "sampling/sampling_logp_difference/mean": 0.011591816507279872,
+ "step": 610,
+ "step_time": 10.539130839891731
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1739.8,
+ "completions/max_terminated_length": 1545.4,
+ "completions/mean_length": 1192.700048828125,
+ "completions/mean_terminated_length": 1113.4033569335938,
+ "completions/min_length": 698.0,
+ "completions/min_terminated_length": 698.0,
+ "entropy": 0.1430244820813338,
+ "epoch": 0.22404371584699453,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.008428337052464485,
+ "kl": 0.00040708660574940346,
+ "learning_rate": 4.0855061349693246e-07,
+ "loss": 0.0004372816067188978,
+ "num_tokens": 4030967.0,
+ "reward": 2.020000123977661,
+ "reward_std": 0.8252323269844055,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.4666666746139526,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.5533333420753479,
+ "rewards/multi_component_reward/std": 0.4648365914821625,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.06130031887441874,
+ "sampling/importance_sampling_ratio/mean": 0.01900877393782139,
+ "sampling/importance_sampling_ratio/min": 0.00034280085452338406,
+ "sampling/sampling_logp_difference/max": 0.37778170108795167,
+ "sampling/sampling_logp_difference/mean": 0.009492585808038712,
+ "step": 615,
+ "step_time": 15.555601417645812
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1618.8,
+ "completions/max_terminated_length": 932.0,
+ "completions/mean_length": 796.36669921875,
+ "completions/mean_terminated_length": 658.6666809082031,
+ "completions/min_length": 447.4,
+ "completions/min_terminated_length": 447.4,
+ "entropy": 0.24623483493924142,
+ "epoch": 0.22586520947176686,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.031970929354429245,
+ "kl": 0.0006379926751833409,
+ "learning_rate": 4.075920245398773e-07,
+ "loss": -0.0030921820551157,
+ "num_tokens": 4057612.0,
+ "reward": 1.3433334350585937,
+ "reward_std": 0.5004158258438111,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.20655910968780516,
+ "rewards/multi_component_reward/mean": 0.1766666740179062,
+ "rewards/multi_component_reward/std": 0.33977961242198945,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.06918634744361044,
+ "sampling/importance_sampling_ratio/mean": 0.02087077551987022,
+ "sampling/importance_sampling_ratio/min": 0.0005975837658579486,
+ "sampling/sampling_logp_difference/max": 0.34571409225463867,
+ "sampling/sampling_logp_difference/mean": 0.01637477772310376,
+ "step": 620,
+ "step_time": 13.71192037537694
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1644.8,
+ "completions/max_terminated_length": 1458.6,
+ "completions/mean_length": 1135.6000244140625,
+ "completions/mean_terminated_length": 976.4267211914063,
+ "completions/min_length": 573.8,
+ "completions/min_terminated_length": 573.8,
+ "entropy": 0.15917381308972836,
+ "epoch": 0.22768670309653916,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.03386072441935539,
+ "kl": 0.0005424051094450987,
+ "learning_rate": 4.066334355828221e-07,
+ "loss": 0.0008449406363070011,
+ "num_tokens": 4094980.0,
+ "reward": 1.4666667938232423,
+ "reward_std": 0.5768125057220459,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.20655910968780516,
+ "rewards/multi_component_reward/mean": 0.20000001415610313,
+ "rewards/multi_component_reward/std": 0.3994357705116272,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.054127467796206474,
+ "sampling/importance_sampling_ratio/mean": 0.016182091273367404,
+ "sampling/importance_sampling_ratio/min": 0.002016719186418556,
+ "sampling/sampling_logp_difference/max": 0.3768587350845337,
+ "sampling/sampling_logp_difference/mean": 0.009851652942597867,
+ "step": 625,
+ "step_time": 14.74142508301884
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1218.6,
+ "completions/max_terminated_length": 1002.4,
+ "completions/mean_length": 759.3666870117188,
+ "completions/mean_terminated_length": 716.2800170898438,
+ "completions/min_length": 491.2,
+ "completions/min_terminated_length": 491.2,
+ "entropy": 0.19249504134058953,
+ "epoch": 0.22950819672131148,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.040883224457502365,
+ "kl": 0.0008160913091463347,
+ "learning_rate": 4.0567484662576683e-07,
+ "loss": -0.02089705318212509,
+ "num_tokens": 4121865.0,
+ "reward": 1.6600001096725463,
+ "reward_std": 0.8352818369865418,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.2666666716337204,
+ "rewards/correctness/std": 0.3977532863616943,
+ "rewards/multi_component_reward/mean": 0.3933333531022072,
+ "rewards/multi_component_reward/std": 0.4375286281108856,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.12170692235231399,
+ "sampling/importance_sampling_ratio/mean": 0.031014200672507285,
+ "sampling/importance_sampling_ratio/min": 0.001588601156254299,
+ "sampling/sampling_logp_difference/max": 0.3645863771438599,
+ "sampling/sampling_logp_difference/mean": 0.01216810904443264,
+ "step": 630,
+ "step_time": 11.676486384868621
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.0,
+ "completions/max_length": 1104.8,
+ "completions/max_terminated_length": 1104.8,
+ "completions/mean_length": 809.733349609375,
+ "completions/mean_terminated_length": 809.733349609375,
+ "completions/min_length": 576.0,
+ "completions/min_terminated_length": 576.0,
+ "entropy": 0.17598983322580655,
+ "epoch": 0.23132969034608378,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.016296500340104103,
+ "kl": 0.0005092226871056482,
+ "learning_rate": 4.0471625766871167e-07,
+ "loss": -0.004767249897122383,
+ "num_tokens": 4148941.0,
+ "reward": 1.520000123977661,
+ "reward_std": 0.6183947920799255,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.3200000077486038,
+ "rewards/multi_component_reward/std": 0.3239211142063141,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.17697620438411832,
+ "sampling/importance_sampling_ratio/mean": 0.06787280989810825,
+ "sampling/importance_sampling_ratio/min": 0.03228051500591391,
+ "sampling/sampling_logp_difference/max": 0.34911930561065674,
+ "sampling/sampling_logp_difference/mean": 0.011234280280768871,
+ "step": 635,
+ "step_time": 10.972807624004782
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1261.6,
+ "completions/max_terminated_length": 1218.8,
+ "completions/mean_length": 815.0000244140625,
+ "completions/mean_terminated_length": 784.780029296875,
+ "completions/min_length": 532.2,
+ "completions/min_terminated_length": 532.2,
+ "entropy": 0.14789476320147515,
+ "epoch": 0.2331511839708561,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.08125295490026474,
+ "kl": 0.0006361469063752641,
+ "learning_rate": 4.037576687116564e-07,
+ "loss": -0.008612464368343353,
+ "num_tokens": 4176637.0,
+ "reward": 1.7100001096725463,
+ "reward_std": 0.8422193884849548,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.3000000089406967,
+ "rewards/correctness/std": 0.3914883255958557,
+ "rewards/multi_component_reward/mean": 0.41000000983476637,
+ "rewards/multi_component_reward/std": 0.4542399704456329,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.14553753491491078,
+ "sampling/importance_sampling_ratio/mean": 0.0640607855282724,
+ "sampling/importance_sampling_ratio/min": 0.0038686898650384326,
+ "sampling/sampling_logp_difference/max": 0.3177733182907104,
+ "sampling/sampling_logp_difference/mean": 0.00996503084897995,
+ "step": 640,
+ "step_time": 11.762126618809997
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1707.4,
+ "completions/max_terminated_length": 1151.8,
+ "completions/mean_length": 1026.000048828125,
+ "completions/mean_terminated_length": 833.5666870117187,
+ "completions/min_length": 601.6,
+ "completions/min_terminated_length": 601.6,
+ "entropy": 0.15211746419469516,
+ "epoch": 0.23497267759562843,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.04184797778725624,
+ "kl": 0.0004896697394239406,
+ "learning_rate": 4.027990797546012e-07,
+ "loss": -0.0011240935884416104,
+ "num_tokens": 4210123.0,
+ "reward": 1.7233334302902221,
+ "reward_std": 0.8806208968162537,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.33333333432674406,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.35666669607162477,
+ "rewards/multi_component_reward/std": 0.5503794223070144,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.10542454989627004,
+ "sampling/importance_sampling_ratio/mean": 0.03725480753928423,
+ "sampling/importance_sampling_ratio/min": 0.0006887080645215871,
+ "sampling/sampling_logp_difference/max": 0.3569879293441772,
+ "sampling/sampling_logp_difference/mean": 0.011003307159990072,
+ "step": 645,
+ "step_time": 14.911480633541942
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1574.6,
+ "completions/max_terminated_length": 1237.0,
+ "completions/mean_length": 920.3000244140625,
+ "completions/mean_terminated_length": 845.8533569335938,
+ "completions/min_length": 600.0,
+ "completions/min_terminated_length": 600.0,
+ "entropy": 0.2023964905490478,
+ "epoch": 0.23679417122040072,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.023950839415192604,
+ "kl": 0.0005954526874120347,
+ "learning_rate": 4.01840490797546e-07,
+ "loss": 0.0016510456800460816,
+ "num_tokens": 4241086.0,
+ "reward": 1.4766667366027832,
+ "reward_std": 0.5016432344913483,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2128240704536438,
+ "rewards/multi_component_reward/mean": 0.24333334639668464,
+ "rewards/multi_component_reward/std": 0.29857338666915895,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.02175317257642746,
+ "sampling/importance_sampling_ratio/mean": 0.007160785421729088,
+ "sampling/importance_sampling_ratio/min": 0.00033338346147502305,
+ "sampling/sampling_logp_difference/max": 0.387990665435791,
+ "sampling/sampling_logp_difference/mean": 0.012314715608954429,
+ "step": 650,
+ "step_time": 13.970132818445563
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.03333333432674408,
+ "completions/max_length": 1193.4,
+ "completions/max_terminated_length": 1015.4,
+ "completions/mean_length": 716.5000244140625,
+ "completions/mean_terminated_length": 674.5733520507813,
+ "completions/min_length": 464.6,
+ "completions/min_terminated_length": 464.6,
+ "entropy": 0.16268871029218038,
+ "epoch": 0.23861566484517305,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.06329948455095291,
+ "kl": 0.0005586101295193657,
+ "learning_rate": 4.0088190184049077e-07,
+ "loss": -0.0020235760137438776,
+ "num_tokens": 4265005.0,
+ "reward": 1.5000001192092896,
+ "reward_std": 0.7927483677864074,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.30000000447034836,
+ "rewards/multi_component_reward/std": 0.4276039183139801,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.08887637332081795,
+ "sampling/importance_sampling_ratio/mean": 0.04327146988362074,
+ "sampling/importance_sampling_ratio/min": 0.01085926350569757,
+ "sampling/sampling_logp_difference/max": 0.34261035919189453,
+ "sampling/sampling_logp_difference/mean": 0.010840693116188049,
+ "step": 655,
+ "step_time": 11.254522551037372
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1448.4,
+ "completions/max_terminated_length": 1037.6,
+ "completions/mean_length": 916.8000366210938,
+ "completions/mean_terminated_length": 796.9666870117187,
+ "completions/min_length": 576.2,
+ "completions/min_terminated_length": 576.2,
+ "entropy": 0.1381769967575868,
+ "epoch": 0.24043715846994534,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.018603617325425148,
+ "kl": 0.0004810667795633587,
+ "learning_rate": 3.9992331288343556e-07,
+ "loss": -0.00833008587360382,
+ "num_tokens": 4294873.0,
+ "reward": 1.530000126361847,
+ "reward_std": 0.5924062907695771,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.23333333134651185,
+ "rewards/correctness/std": 0.24494898319244385,
+ "rewards/multi_component_reward/mean": 0.29666669070720675,
+ "rewards/multi_component_reward/std": 0.35217125415802003,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.19538256973028184,
+ "sampling/importance_sampling_ratio/mean": 0.0801917064934969,
+ "sampling/importance_sampling_ratio/min": 0.002801872471235356,
+ "sampling/sampling_logp_difference/max": 0.36407305002212526,
+ "sampling/sampling_logp_difference/mean": 0.009397336840629577,
+ "step": 660,
+ "step_time": 13.188869021646678
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.0,
+ "completions/max_length": 915.0,
+ "completions/max_terminated_length": 915.0,
+ "completions/mean_length": 611.6000183105468,
+ "completions/mean_terminated_length": 611.6000183105468,
+ "completions/min_length": 423.0,
+ "completions/min_terminated_length": 423.0,
+ "entropy": 0.21465270320574442,
+ "epoch": 0.24225865209471767,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.021906884387135506,
+ "kl": 0.0005874778813449666,
+ "learning_rate": 3.9896472392638035e-07,
+ "loss": 0.0034822095185518265,
+ "num_tokens": 4315237.0,
+ "reward": 1.9400000810623168,
+ "reward_std": 0.8352818369865418,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.4000000089406967,
+ "rewards/correctness/std": 0.3977532863616943,
+ "rewards/multi_component_reward/mean": 0.5399999991059303,
+ "rewards/multi_component_reward/std": 0.4375286281108856,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.08231882937252522,
+ "sampling/importance_sampling_ratio/mean": 0.03466434141155332,
+ "sampling/importance_sampling_ratio/min": 0.0024839662786689588,
+ "sampling/sampling_logp_difference/max": 0.35751450061798096,
+ "sampling/sampling_logp_difference/mean": 0.013147631753236055,
+ "step": 665,
+ "step_time": 9.326043193787337
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1493.4,
+ "completions/max_terminated_length": 1139.0,
+ "completions/mean_length": 914.633349609375,
+ "completions/mean_terminated_length": 779.0366821289062,
+ "completions/min_length": 560.8,
+ "completions/min_terminated_length": 560.8,
+ "entropy": 0.15737735082705814,
+ "epoch": 0.24408014571949,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.01330585964024067,
+ "kl": 0.0005065599815376724,
+ "learning_rate": 3.9800613496932514e-07,
+ "loss": -0.006443501263856888,
+ "num_tokens": 4345802.0,
+ "reward": 1.32000013589859,
+ "reward_std": 0.6418773233890533,
+ "rewards/boxed_rate/mean": 0.9,
+ "rewards/boxed_rate/std": 0.18492921590805053,
+ "rewards/correctness/mean": 0.13333333730697633,
+ "rewards/correctness/std": 0.2665788769721985,
+ "rewards/multi_component_reward/mean": 0.18666667491197586,
+ "rewards/multi_component_reward/std": 0.3788073122501373,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.18492921590805053,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.10360531127080322,
+ "sampling/importance_sampling_ratio/mean": 0.05636316137388349,
+ "sampling/importance_sampling_ratio/min": 0.022174429874797168,
+ "sampling/sampling_logp_difference/max": 0.37078907489776614,
+ "sampling/sampling_logp_difference/mean": 0.010791171807795763,
+ "step": 670,
+ "step_time": 13.329693194106222
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1439.8,
+ "completions/max_terminated_length": 1202.4,
+ "completions/mean_length": 803.2666870117188,
+ "completions/mean_terminated_length": 723.4000122070313,
+ "completions/min_length": 409.0,
+ "completions/min_terminated_length": 409.0,
+ "entropy": 0.17263933196663855,
+ "epoch": 0.2459016393442623,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.1460752934217453,
+ "kl": 0.0005690245316751923,
+ "learning_rate": 3.970475460122699e-07,
+ "loss": -0.012128566205501557,
+ "num_tokens": 4373710.0,
+ "reward": 1.6166667461395263,
+ "reward_std": 0.8088707447052002,
+ "rewards/boxed_rate/mean": 0.9666666626930237,
+ "rewards/boxed_rate/std": 0.08164966106414795,
+ "rewards/correctness/mean": 0.23333334028720856,
+ "rewards/correctness/std": 0.3914883255958557,
+ "rewards/multi_component_reward/mean": 0.31666667610406873,
+ "rewards/multi_component_reward/std": 0.4805600345134735,
+ "rewards/no_answer_rate/mean": 0.03333333432674408,
+ "rewards/no_answer_rate/std": 0.08164966106414795,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.10961102806031704,
+ "sampling/importance_sampling_ratio/mean": 0.043116622418165204,
+ "sampling/importance_sampling_ratio/min": 0.001660405399086784,
+ "sampling/sampling_logp_difference/max": 0.3272708892822266,
+ "sampling/sampling_logp_difference/mean": 0.011050505377352238,
+ "step": 675,
+ "step_time": 12.905224220454693
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333333432674408,
+ "completions/max_length": 1583.4,
+ "completions/max_terminated_length": 1101.4,
+ "completions/mean_length": 1081.7000366210937,
+ "completions/mean_terminated_length": 832.26669921875,
+ "completions/min_length": 648.0,
+ "completions/min_terminated_length": 648.0,
+ "entropy": 0.17594740043083826,
+ "epoch": 0.24772313296903462,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.07136747241020203,
+ "kl": 0.0005246107791511653,
+ "learning_rate": 3.960889570552147e-07,
+ "loss": 0.003443557769060135,
+ "num_tokens": 4409875.0,
+ "reward": 1.4266667485237121,
+ "reward_std": 0.781731303036213,
+ "rewards/boxed_rate/mean": 0.7666666626930236,
+ "rewards/boxed_rate/std": 0.30073869228363037,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.16000001206994058,
+ "rewards/multi_component_reward/std": 0.509483140707016,
+ "rewards/no_answer_rate/mean": 0.23333333432674408,
+ "rewards/no_answer_rate/std": 0.30073869228363037,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.07207991853356362,
+ "sampling/importance_sampling_ratio/mean": 0.018585072737187146,
+ "sampling/importance_sampling_ratio/min": 0.00033181315377639464,
+ "sampling/sampling_logp_difference/max": 0.35654211044311523,
+ "sampling/sampling_logp_difference/mean": 0.011251153517514467,
+ "step": 680,
+ "step_time": 14.236122718639672
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.23333334028720856,
+ "completions/max_length": 1406.2,
+ "completions/max_terminated_length": 905.2,
+ "completions/mean_length": 937.7333770751953,
+ "completions/mean_terminated_length": 666.7800079345703,
+ "completions/min_length": 499.4,
+ "completions/min_terminated_length": 499.4,
+ "entropy": 0.15866130826373895,
+ "epoch": 0.2495446265938069,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.019822267815470695,
+ "kl": 0.0025144809878838714,
+ "learning_rate": 3.9513036809815945e-07,
+ "loss": 0.0014299599453806878,
+ "num_tokens": 4441367.0,
+ "reward": 1.5433334589004517,
+ "reward_std": 0.8197273433208465,
+ "rewards/boxed_rate/mean": 0.7666666686534882,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.26666666865348815,
+ "rewards/correctness/std": 0.32236858606338503,
+ "rewards/multi_component_reward/mean": 0.2433333471417427,
+ "rewards/multi_component_reward/std": 0.5061476826667786,
+ "rewards/no_answer_rate/mean": 0.23333334028720856,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.19941046759486197,
+ "sampling/importance_sampling_ratio/mean": 0.10512657705694436,
+ "sampling/importance_sampling_ratio/min": 0.052634853626081814,
+ "sampling/sampling_logp_difference/max": 0.39468674659729003,
+ "sampling/sampling_logp_difference/mean": 0.010894744656980038,
+ "step": 685,
+ "step_time": 13.195306334272027
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1319.4,
+ "completions/max_terminated_length": 1112.6,
+ "completions/mean_length": 827.1333801269532,
+ "completions/mean_terminated_length": 760.5666931152343,
+ "completions/min_length": 536.4,
+ "completions/min_terminated_length": 536.4,
+ "entropy": 0.18860391999284426,
+ "epoch": 0.25136612021857924,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.028540268540382385,
+ "kl": 0.0006184172428523501,
+ "learning_rate": 3.941717791411043e-07,
+ "loss": -0.008557839691638947,
+ "num_tokens": 4468635.0,
+ "reward": 2.013333463668823,
+ "reward_std": 0.8281045794487,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.4333333343267441,
+ "rewards/correctness/std": 0.3761233925819397,
+ "rewards/multi_component_reward/mean": 0.513333360850811,
+ "rewards/multi_component_reward/std": 0.4702867269515991,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.06727873869240283,
+ "sampling/importance_sampling_ratio/mean": 0.03433487480506301,
+ "sampling/importance_sampling_ratio/min": 0.00048347263269847286,
+ "sampling/sampling_logp_difference/max": 0.42853312492370604,
+ "sampling/sampling_logp_difference/mean": 0.012394568137824535,
+ "step": 690,
+ "step_time": 12.268561779148877
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1604.2,
+ "completions/max_terminated_length": 1084.2,
+ "completions/mean_length": 907.400048828125,
+ "completions/mean_terminated_length": 778.053369140625,
+ "completions/min_length": 563.2,
+ "completions/min_terminated_length": 563.2,
+ "entropy": 0.14893935720125834,
+ "epoch": 0.25318761384335153,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.018119433894753456,
+ "kl": 0.00045477717115621394,
+ "learning_rate": 3.932131901840491e-07,
+ "loss": -0.002763097546994686,
+ "num_tokens": 4499319.0,
+ "reward": 1.3900001049041748,
+ "reward_std": 0.6540532588958741,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.16666666865348817,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.2233333483338356,
+ "rewards/multi_component_reward/std": 0.3906372755765915,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.09045649766921997,
+ "sampling/importance_sampling_ratio/mean": 0.034694204293191436,
+ "sampling/importance_sampling_ratio/min": 0.0024610218745878347,
+ "sampling/sampling_logp_difference/max": 0.3914134740829468,
+ "sampling/sampling_logp_difference/mean": 0.010633220244199038,
+ "step": 695,
+ "step_time": 13.952417024224996
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1497.8,
+ "completions/max_terminated_length": 771.4,
+ "completions/mean_length": 722.333349609375,
+ "completions/mean_terminated_length": 584.6333435058593,
+ "completions/min_length": 428.8,
+ "completions/min_terminated_length": 428.8,
+ "entropy": 0.20499998504916828,
+ "epoch": 0.2550091074681239,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.06937753409147263,
+ "kl": 0.0006178093899507076,
+ "learning_rate": 3.922546012269938e-07,
+ "loss": -0.01590590924024582,
+ "num_tokens": 4523047.0,
+ "reward": 1.4600000619888305,
+ "reward_std": 0.7687346935272217,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.20000000596046447,
+ "rewards/correctness/std": 0.3098386645317078,
+ "rewards/multi_component_reward/mean": 0.2600000202655792,
+ "rewards/multi_component_reward/std": 0.4624049305915833,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.16114517226815223,
+ "sampling/importance_sampling_ratio/mean": 0.06391621977090836,
+ "sampling/importance_sampling_ratio/min": 0.010175443676988779,
+ "sampling/sampling_logp_difference/max": 0.3058194637298584,
+ "sampling/sampling_logp_difference/mean": 0.013935025222599507,
+ "step": 700,
+ "step_time": 13.013086147420108
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.16666666865348817,
+ "completions/max_length": 1429.0,
+ "completions/max_terminated_length": 942.0,
+ "completions/mean_length": 947.1667053222657,
+ "completions/mean_terminated_length": 811.6733581542969,
+ "completions/min_length": 627.2,
+ "completions/min_terminated_length": 627.2,
+ "entropy": 0.17254207134246827,
+ "epoch": 0.2568306010928962,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.012989591807126999,
+ "kl": 0.0013695822664885782,
+ "learning_rate": 3.9129601226993866e-07,
+ "loss": 0.02462577521800995,
+ "num_tokens": 4554762.0,
+ "reward": 2.000000071525574,
+ "reward_std": 1.0529863953590393,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.4666666626930237,
+ "rewards/correctness/std": 0.4298781991004944,
+ "rewards/multi_component_reward/mean": 0.5000000357627868,
+ "rewards/multi_component_reward/std": 0.5574271440505981,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.1705483404919505,
+ "sampling/importance_sampling_ratio/mean": 0.062803098349832,
+ "sampling/importance_sampling_ratio/min": 0.011600531370969901,
+ "sampling/sampling_logp_difference/max": 0.36405808925628663,
+ "sampling/sampling_logp_difference/mean": 0.01186208426952362,
+ "step": 705,
+ "step_time": 13.669213125482202
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1826.4,
+ "completions/max_terminated_length": 1155.4,
+ "completions/mean_length": 1059.6667114257812,
+ "completions/mean_terminated_length": 841.7533569335938,
+ "completions/min_length": 515.2,
+ "completions/min_terminated_length": 515.2,
+ "entropy": 0.23051666493217152,
+ "epoch": 0.2586520947176685,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.04497559741139412,
+ "kl": 0.000664207028845946,
+ "learning_rate": 3.9033742331288344e-07,
+ "loss": -0.00020618487615138293,
+ "num_tokens": 4589378.0,
+ "reward": 1.8600000977516173,
+ "reward_std": 0.9444810330867768,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.40000000298023225,
+ "rewards/correctness/std": 0.36985843181610106,
+ "rewards/multi_component_reward/mean": 0.4266666829586029,
+ "rewards/multi_component_reward/std": 0.552643096446991,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.04061034563928843,
+ "sampling/importance_sampling_ratio/mean": 0.008759467327035964,
+ "sampling/importance_sampling_ratio/min": 0.0002094253270666409,
+ "sampling/sampling_logp_difference/max": 0.3215909957885742,
+ "sampling/sampling_logp_difference/mean": 0.014831000193953513,
+ "step": 710,
+ "step_time": 15.66096483822912
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.06666666865348816,
+ "completions/max_length": 1843.8,
+ "completions/max_terminated_length": 1615.2,
+ "completions/mean_length": 1085.3667114257812,
+ "completions/mean_terminated_length": 1008.9400390625,
+ "completions/min_length": 637.2,
+ "completions/min_terminated_length": 637.2,
+ "entropy": 0.19345977554718655,
+ "epoch": 0.2604735883424408,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.04418517276644707,
+ "kl": 0.0008112916897516697,
+ "learning_rate": 3.893788343558282e-07,
+ "loss": -0.0014110441319644452,
+ "num_tokens": 4626127.0,
+ "reward": 1.5000000953674317,
+ "reward_std": 0.7173020958900451,
+ "rewards/boxed_rate/mean": 0.9333333253860474,
+ "rewards/boxed_rate/std": 0.1632993221282959,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.29447373151779177,
+ "rewards/multi_component_reward/mean": 0.266666679084301,
+ "rewards/multi_component_reward/std": 0.3763957768678665,
+ "rewards/no_answer_rate/mean": 0.06666666865348816,
+ "rewards/no_answer_rate/std": 0.1632993221282959,
+ "rewards/repetition_rate/mean": 0.03333333432674408,
+ "rewards/repetition_rate/std": 0.08164966106414795,
+ "sampling/importance_sampling_ratio/max": 0.04309340715408325,
+ "sampling/importance_sampling_ratio/mean": 0.012624496826902032,
+ "sampling/importance_sampling_ratio/min": 0.00014569541402083248,
+ "sampling/sampling_logp_difference/max": 0.3676969051361084,
+ "sampling/sampling_logp_difference/mean": 0.01275388365611434,
+ "step": 715,
+ "step_time": 15.848505285196007
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.1666666716337204,
+ "completions/max_length": 1794.8,
+ "completions/max_terminated_length": 1234.6,
+ "completions/mean_length": 973.4000183105469,
+ "completions/mean_terminated_length": 763.0933410644532,
+ "completions/min_length": 480.8,
+ "completions/min_terminated_length": 480.8,
+ "entropy": 0.19203649039069812,
+ "epoch": 0.26229508196721313,
+ "frac_reward_zero_std": 0.2,
+ "grad_norm": 0.01788559928536415,
+ "kl": 0.0007585312666681906,
+ "learning_rate": 3.8842024539877297e-07,
+ "loss": -0.0016064226627349853,
+ "num_tokens": 4659067.0,
+ "reward": 1.3966667771339416,
+ "reward_std": 0.3670642614364624,
+ "rewards/boxed_rate/mean": 0.8333333373069763,
+ "rewards/boxed_rate/std": 0.2882087707519531,
+ "rewards/correctness/mean": 0.1666666626930237,
+ "rewards/correctness/std": 0.08164966106414795,
+ "rewards/multi_component_reward/mean": 0.16333335414528846,
+ "rewards/multi_component_reward/std": 0.28173157572746277,
+ "rewards/no_answer_rate/mean": 0.1666666716337204,
+ "rewards/no_answer_rate/std": 0.2882087707519531,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.1632993221282959,
+ "sampling/importance_sampling_ratio/max": 0.2071637876331806,
+ "sampling/importance_sampling_ratio/mean": 0.051915486436337234,
+ "sampling/importance_sampling_ratio/min": 3.873565276598517e-06,
+ "sampling/sampling_logp_difference/max": 0.34898459911346436,
+ "sampling/sampling_logp_difference/mean": 0.013108029030263424,
+ "step": 720,
+ "step_time": 15.256685120798647
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000596046447,
+ "completions/max_length": 1793.6,
+ "completions/max_terminated_length": 1117.6,
+ "completions/mean_length": 1055.9667114257813,
+ "completions/mean_terminated_length": 758.4666870117187,
+ "completions/min_length": 497.0,
+ "completions/min_terminated_length": 497.0,
+ "entropy": 0.20671888664364815,
+ "epoch": 0.2641165755919854,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 0.023016415536403656,
+ "kl": 0.0006531716018798762,
+ "learning_rate": 3.874616564417178e-07,
+ "loss": 0.0004923004657030105,
+ "num_tokens": 4692996.0,
+ "reward": 1.6100001096725465,
+ "reward_std": 0.5153546214103699,
+ "rewards/boxed_rate/mean": 0.7999999940395355,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.3000000029802322,
+ "rewards/correctness/std": 0.18492921590805053,
+ "rewards/multi_component_reward/mean": 0.310000017285347,
+ "rewards/multi_component_reward/std": 0.3351393073797226,
+ "rewards/no_answer_rate/mean": 0.20000000596046447,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.050453129876405,
+ "sampling/importance_sampling_ratio/mean": 0.014345021429471671,
+ "sampling/importance_sampling_ratio/min": 0.0007461899496814892,
+ "sampling/sampling_logp_difference/max": 0.4133040189743042,
+ "sampling/sampling_logp_difference/mean": 0.012775494437664747,
+ "step": 725,
+ "step_time": 15.257122083753348
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.20000000298023224,
+ "completions/max_length": 1528.6,
+ "completions/max_terminated_length": 1156.6,
+ "completions/mean_length": 1054.86669921875,
+ "completions/mean_terminated_length": 859.533349609375,
+ "completions/min_length": 625.0,
+ "completions/min_terminated_length": 625.0,
+ "entropy": 0.21140047113100688,
+ "epoch": 0.2659380692167577,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.018997710198163986,
+ "kl": 0.0006346999696688726,
+ "learning_rate": 3.8650306748466255e-07,
+ "loss": 0.0016422737389802933,
+ "num_tokens": 4727378.0,
+ "reward": 1.5333333969116212,
+ "reward_std": 0.7559072077274323,
+ "rewards/boxed_rate/mean": 0.8333333253860473,
+ "rewards/boxed_rate/std": 0.2728438377380371,
+ "rewards/correctness/mean": 0.23333333432674408,
+ "rewards/correctness/std": 0.30073869228363037,
+ "rewards/multi_component_reward/mean": 0.23333334922790527,
+ "rewards/multi_component_reward/std": 0.4953298598527908,
+ "rewards/no_answer_rate/mean": 0.16666666865348817,
+ "rewards/no_answer_rate/std": 0.2728438377380371,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.08731604730710388,
+ "sampling/importance_sampling_ratio/mean": 0.020629775966517626,
+ "sampling/importance_sampling_ratio/min": 0.0005315004007400171,
+ "sampling/sampling_logp_difference/max": 0.3350215911865234,
+ "sampling/sampling_logp_difference/mean": 0.013690494932234287,
+ "step": 730,
+ "step_time": 13.985034779645503
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.13333333730697633,
+ "completions/max_length": 1815.2,
+ "completions/max_terminated_length": 1340.4,
+ "completions/mean_length": 1031.46669921875,
+ "completions/mean_terminated_length": 883.0133483886718,
+ "completions/min_length": 518.6,
+ "completions/min_terminated_length": 518.6,
+ "entropy": 0.18139945759127538,
+ "epoch": 0.2677595628415301,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.021211547777056694,
+ "kl": 0.000629005294952852,
+ "learning_rate": 3.8554447852760733e-07,
+ "loss": 0.0014220052398741245,
+ "num_tokens": 4761916.0,
+ "reward": 1.4866667628288268,
+ "reward_std": 0.8185165703296662,
+ "rewards/boxed_rate/mean": 0.8666666626930237,
+ "rewards/boxed_rate/std": 0.2665788769721985,
+ "rewards/correctness/mean": 0.20000000298023224,
+ "rewards/correctness/std": 0.35449349880218506,
+ "rewards/multi_component_reward/mean": 0.2200000137090683,
+ "rewards/multi_component_reward/std": 0.5127273321151733,
+ "rewards/no_answer_rate/mean": 0.13333333730697633,
+ "rewards/no_answer_rate/std": 0.2665788769721985,
+ "rewards/repetition_rate/mean": 0.06666666865348816,
+ "rewards/repetition_rate/std": 0.10327955484390258,
+ "sampling/importance_sampling_ratio/max": 0.06554706878960133,
+ "sampling/importance_sampling_ratio/mean": 0.02508344785310328,
+ "sampling/importance_sampling_ratio/min": 0.0035423538342001966,
+ "sampling/sampling_logp_difference/max": 0.31146485805511476,
+ "sampling/sampling_logp_difference/mean": 0.0113022543489933,
+ "step": 735,
+ "step_time": 15.48011973593384
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.0,
+ "completions/max_length": 883.8,
+ "completions/max_terminated_length": 883.8,
+ "completions/mean_length": 542.2000122070312,
+ "completions/mean_terminated_length": 542.2000122070312,
+ "completions/min_length": 352.4,
+ "completions/min_terminated_length": 352.4,
+ "entropy": 0.19473386829098066,
+ "epoch": 0.26958105646630237,
+ "frac_reward_zero_std": 0.4,
+ "grad_norm": 2.118746042251587,
+ "kl": 0.0021321156071887042,
+ "learning_rate": 3.845858895705521e-07,
+ "loss": -0.0064903564751148226,
+ "num_tokens": 4781002.0,
+ "reward": 1.7300001382827759,
+ "reward_std": 0.5729720234870911,
+ "rewards/boxed_rate/mean": 1.0,
+ "rewards/boxed_rate/std": 0.0,
+ "rewards/correctness/mean": 0.29999999701976776,
+ "rewards/correctness/std": 0.2728438377380371,
+ "rewards/multi_component_reward/mean": 0.430000028014183,
+ "rewards/multi_component_reward/std": 0.3001282274723053,
+ "rewards/no_answer_rate/mean": 0.0,
+ "rewards/no_answer_rate/std": 0.0,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.28090705554932355,
+ "sampling/importance_sampling_ratio/mean": 0.15755705069750547,
+ "sampling/importance_sampling_ratio/min": 0.0855626948821282,
+ "sampling/sampling_logp_difference/max": 0.3150986671447754,
+ "sampling/sampling_logp_difference/mean": 0.01262940764427185,
+ "step": 740,
+ "step_time": 9.168580058775841
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.10000000298023223,
+ "completions/max_length": 1972.8,
+ "completions/max_terminated_length": 1486.4,
+ "completions/mean_length": 1172.000048828125,
+ "completions/mean_terminated_length": 1050.6600280761718,
+ "completions/min_length": 720.8,
+ "completions/min_terminated_length": 720.8,
+ "entropy": 0.11767480584482352,
+ "epoch": 0.27140255009107467,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.08145014941692352,
+ "kl": 0.0006028921416145749,
+ "learning_rate": 3.836273006134969e-07,
+ "loss": 0.0025216279551386832,
+ "num_tokens": 4818496.0,
+ "reward": 1.6700001001358031,
+ "reward_std": 0.8164093613624572,
+ "rewards/boxed_rate/mean": 0.899999988079071,
+ "rewards/boxed_rate/std": 0.24494898319244385,
+ "rewards/correctness/mean": 0.3,
+ "rewards/correctness/std": 0.34822853803634646,
+ "rewards/multi_component_reward/mean": 0.37000002562999723,
+ "rewards/multi_component_reward/std": 0.47640362083911897,
+ "rewards/no_answer_rate/mean": 0.10000000298023223,
+ "rewards/no_answer_rate/std": 0.24494898319244385,
+ "rewards/repetition_rate/mean": 0.0,
+ "rewards/repetition_rate/std": 0.0,
+ "sampling/importance_sampling_ratio/max": 0.14499615281820297,
+ "sampling/importance_sampling_ratio/mean": 0.04840177483856678,
+ "sampling/importance_sampling_ratio/min": 0.0013758533679049771,
+ "sampling/sampling_logp_difference/max": 0.32971014976501467,
+ "sampling/sampling_logp_difference/mean": 0.008220831863582134,
+ "step": 745,
+ "step_time": 16.791488209925593
+ },
+ {
+ "clip_ratio/high_max": 0.0,
+ "clip_ratio/high_mean": 0.0,
+ "clip_ratio/low_mean": 0.0,
+ "clip_ratio/low_min": 0.0,
+ "clip_ratio/region_mean": 0.0,
+ "completions/clipped_ratio": 0.36666667759418486,
+ "completions/max_length": 1719.0,
+ "completions/max_terminated_length": 1232.4,
+ "completions/mean_length": 1217.600048828125,
+ "completions/mean_terminated_length": 822.3233520507813,
+ "completions/min_length": 564.6,
+ "completions/min_terminated_length": 564.6,
+ "entropy": 0.19243225703636804,
+ "epoch": 0.273224043715847,
+ "frac_reward_zero_std": 0.0,
+ "grad_norm": 0.03210171312093735,
+ "kl": 0.0007616788071269791,
+ "learning_rate": 3.826687116564417e-07,
+ "loss": -0.005124400556087494,
+ "num_tokens": 4864732.0,
+ "reward": 1.7300000667572022,
+ "reward_std": 0.9617222905158996,
+ "rewards/boxed_rate/mean": 0.7666666626930236,
+ "rewards/boxed_rate/std": 0.3761233925819397,
+ "rewards/correctness/mean": 0.33333334028720857,
+ "rewards/correctness/std": 0.3977532863616943,
+ "rewards/multi_component_reward/mean": 0.29666668772697447,
+ "rewards/multi_component_reward/std": 0.6426073163747787,
+ "rewards/no_answer_rate/mean": 0.2333333373069763,
+ "rewards/no_answer_rate/std": 0.3761233925819397,
+ "rewards/repetition_rate/mean": 0.1,
+ "rewards/repetition_rate/std": 0.10954451560974121,
+ "sampling/importance_sampling_ratio/max": 0.11619223132729531,
+ "sampling/importance_sampling_ratio/mean": 0.03387247350765392,
+ "sampling/importance_sampling_ratio/min": 0.0005780367053725346,
+ "sampling/sampling_logp_difference/max": 0.3284996747970581,
+ "sampling/sampling_logp_difference/mean": 0.012111451011151075,
+ "step": 750,
+ "step_time": 15.568404103443026
+ },
+ {
+ "epoch": 0.273224043715847,
+ "eval_clip_ratio/high_max": 0.0,
+ "eval_clip_ratio/high_mean": 0.0,
+ "eval_clip_ratio/low_mean": 0.0,
+ "eval_clip_ratio/low_min": 0.0,
+ "eval_clip_ratio/region_mean": 0.0,
+ "eval_completions/clipped_ratio": 0.17000000417232514,
+ "eval_completions/max_length": 1756.9,
+ "eval_completions/max_terminated_length": 1208.56,
+ "eval_completions/mean_length": 1074.366694946289,
+ "eval_completions/mean_terminated_length": 874.3350158691406,
+ "eval_completions/min_length": 579.92,
+ "eval_completions/min_terminated_length": 579.92,
+ "eval_entropy": 0.18485607638955115,
+ "eval_frac_reward_zero_std": 0.1,
+ "eval_kl": 0.0007376758259488269,
+ "eval_loss": -0.0014031080063432455,
+ "eval_num_tokens": 4864732.0,
+ "eval_reward": 1.6413334274291993,
+ "eval_reward_std": 0.7854073830693961,
+ "eval_rewards/boxed_rate/mean": 0.8366666615009308,
+ "eval_rewards/boxed_rate/std": 0.27878632068634035,
+ "eval_rewards/correctness/mean": 0.2933333370089531,
+ "eval_rewards/correctness/std": 0.32616410493850706,
+ "eval_rewards/multi_component_reward/mean": 0.3146666841953993,
+ "eval_rewards/multi_component_reward/std": 0.4994544792175293,
+ "eval_rewards/no_answer_rate/mean": 0.16333333730697633,
+ "eval_rewards/no_answer_rate/std": 0.27878632068634035,
+ "eval_rewards/repetition_rate/mean": 0.03333333432674408,
+ "eval_rewards/repetition_rate/std": 0.06964570760726929,
+ "eval_runtime": 681.9794,
+ "eval_samples_per_second": 0.073,
+ "eval_sampling/importance_sampling_ratio/max": 0.09388977923779748,
+ "eval_sampling/importance_sampling_ratio/mean": 0.029817550851148553,
+ "eval_sampling/importance_sampling_ratio/min": 0.003447885300730625,
+ "eval_sampling/sampling_logp_difference/max": 0.7785206460952758,
+ "eval_sampling/sampling_logp_difference/mean": 0.014042644733563066,
+ "eval_steps_per_second": 0.013,
+ "step": 750
+ }
+ ],
+ "logging_steps": 5,
+ "max_steps": 2745,
+ "num_input_tokens_seen": 4864732,
+ "num_train_epochs": 1,
+ "save_steps": 150,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 0.0,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/training_args.bin b/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..133cb6661b0124bb9db0d7d208d748fd736ea689
--- /dev/null
+++ b/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6696e1fef8bde5e22de2f5ae18c284f2b2d35a345157db5151571518385fb38b
+size 5713