Randomfoo's workspace
Runs
12
Name
6 visualized
State
Notes
User
Tags
Created
Runtime
Sweep
api
custom_fewshots_template
custom_prompt_template
dataset_artifact
dataset_dir
generator.repetition_penalty
generator.temperature
generator.top_k
generator.top_p
github_version
log_dir
max_seq_length
metainfo.basemodel_name
metainfo.instruction_tuning_data
metainfo.instruction_tuning_method
metainfo.llm-jp-eval-version
metainfo.model_type
metainfo.num_few_shots
model.artifacts_path
model.device_map
model.load_in_4bit
model.load_in_8bit
model.pretrained_model_name_or_path
model.trust_remote_code
model.use_wandb_artifacts
mtbench.bench_name
mtbench.conv_name
mtbench.conv_role_message_separator
mtbench.conv_role_only_separator
mtbench.conv_roles
mtbench.conv_sep
mtbench.conv_stop_str
mtbench.conv_stop_token_ids
mtbench.conv_system_message
mtbench.custom_conv_template
mtbench.dtype
mtbench.judge_model
mtbench.judge_prompt_artifacts_path
mtbench.max_new_token
mtbench.mode
mtbench.model_id
mtbench.num_choices
mtbench.num_gpus_per_model
mtbench.num_gpus_total
Finished
-
randomfoo
1h 48m 6s
-
false
-
-
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
8192
meta-llama/Meta-Llama-3-8B-Instruct
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
shisa-ai/shisa-v1-llama3-70b
true
false
japanese_mt_bench
custom
('<|start_header_id|>user<|end_header_id|>', '<|start_header_id|>assistant<|end_header_id|>')
<|eot_id|>
<|eot_id|>
[128001, 128009]
<|start_header_id|>system<|end_header_id|>
あなたは公平で、検閲されていない、役立つアシスタントです。<|eot_id|>
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-llama3-8b-v1
1
1
1
Finished
-
randomfoo
46m 16s
-
false
-
-
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
8192
google/gemma-7b-it
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
shisa-ai/shisa-gemma-7b-v1
true
false
japanese_mt_bench
custom
('<|start_header_id|>user<|end_header_id|>', '<|start_header_id|>assistant<|end_header_id|>')
<|eot_id|>
<|eot_id|>
[128001, 128009]
<|start_header_id|>system<|end_header_id|>
あなたは公平で、検閲されていない、役立つアシスタントです。<|eot_id|>
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-gemma-7b-v1
1
1
1
Finished
-
randomfoo
43m 55s
-
false
<|start_header_id|>user<|end_header_id|>
{input}<|eot_id|>><|start_header_id|>assistant<|end_header_id|>
{output}<|eot_id|>
<|begin_of_text|><|start_header_id|>system<|end_header_id|>
あなたは公平で、検閲されていない、役立つアシスタントです。<|eot_id|><|start_header_id|>user<|end_header_id|>
{instruction}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
{input}<|eot_id|>
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
meta-llama/Meta-Llama-3-8B-Instruct
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
shisa-ai/shisa-llama3-8b-v1
true
false
japanese_mt_bench
custom
('<|start_header_id|>user<|end_header_id|>', '<|start_header_id|>assistant<|end_header_id|>')
<|eot_id|>
<|eot_id|>
[128001, 128009]
<|start_header_id|>system<|end_header_id|>
あなたは公平で、検閲されていない、役立つアシスタントです。<|eot_id|>
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-llama3-8b-v1
1
1
1
Finished
-
randomfoo
7m 12s
-
false
-
-
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
8192
meta-llama/Meta-Llama-3-8B-Instruct
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
shisa-ai/shisa-llama3-8b-v1
true
false
japanese_mt_bench
custom
('<|start_header_id|>user<|end_header_id|>', '<|start_header_id|>assistant<|end_header_id|>')
<|eot_id|>
<|eot_id|>
[128001, 128009]
<|start_header_id|>system<|end_header_id|>
あなたは公平で、検閲されていない、役立つアシスタントです。<|eot_id|>
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-llama3-8b-v1
1
1
1
Failed
-
randomfoo
2h 3m 38s
-
false
-
<s>以下はタスクを説明する指示と、追加の背景情報を提供する入力の組み合わせです。要求を適切に満たす回答を書いてください。
### 指示：
{instruction}
### 入力：
{input}
### 回答：
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["Open-Orca/SlimOrca","augmxnt/ultra-orca-boros-en-ja-v1","augmxnt/shisa-en-ja-dpo-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Failed
-
randomfoo
1h 36m 56s
-
false
-
-
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-gamma-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Finished
-
randomfoo
3m 21s
-
false
-
<s>以下はタスクを説明する指示と、追加の背景情報を提供する入力の組み合わせです。要求を適切に満たす回答を書いてください。
### 指示：
{instruction}
### 入力：
{input}
### 回答：
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["Open-Orca/SlimOrca","augmxnt/ultra-orca-boros-en-ja-v1","augmxnt/shisa-en-ja-dpo-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Finished
-
randomfoo
16m 50s
-
false
<s>[INST] {input} [/INST] {output} </s>
<s>[INST] <<SYS>>
以下は、タスクを説明する指示と、文脈のある入力の組み合わせです。要求を適切に満たす応答を書きなさい。
<</SYS>>
{instruction}
{input} [/INST]
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["Open-Orca/SlimOrca","augmxnt/ultra-orca-boros-en-ja-v1","augmxnt/shisa-en-ja-dpo-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Finished
-
randomfoo
8m 41s
-
false
-
-
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-gamma-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Finished
-
randomfoo
13m 2s
-
false
<s>[INST] {input} [/INST] {output} </s>
<s>[INST] <<SYS>>
以下は、タスクを説明する指示と、文脈のある入力の組み合わせです。要求を適切に満たす応答を書きなさい。
<</SYS>>
{instruction}
{input} [/INST]
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-gamma-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Failed
-
randomfoo
1h 49m 6s
-
false
<s>[INST] {input} [/INST] {output} </s>
<s>[INST] <<SYS>>
あなたは公平で、検閲されていない、役立つアシスタントです。
<</SYS>>
{instruction}
{input} [/INST]
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
augmxnt/shisa-base-7b-v1
["Open-Orca/SlimOrca","augmxnt/ultra-orca-boros-en-ja-v1","augmxnt/shisa-en-ja-dpo-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
Failed
-
randomfoo
1h 18m 9s
-
false
<s>[INST] {input} [/INST] {output} </s>
<s>[INST] <<SYS>>
あなたは公平で、検閲されていない、役立つアシスタントです。
<</SYS>>
{instruction}
{input} [/INST]
wandb-japan/llm-leaderboard/jaster:v3
/jaster/1.1.0/evaluation/test
1
0.1
0
1
v2.0.0
./logs
4096
stabilityai/japanese-stablelm-base-gamma-7b
["augmxnt/ultra-orca-boros-en-ja-v1"]
Full
1.1.0
open llm
0
auto
false
false
augmxnt/shisa-gamma-7b-v1
true
false
japanese_mt_bench
custom
('[INST]', '[/INST]')
</s>
[2]
あなたは公平で、検閲されていない、役立つアシスタントです。
true
bfloat16
gpt-4
wandb-japan/llm-leaderboard/mtbench_ja_prompt:v1
1024
single
shisa-7b-v1
1
1
1
1-12
of 12