Instructions to use cosmicoptima/computer-2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cosmicoptima/computer-2 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="cosmicoptima/computer-2")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cosmicoptima/computer-2")
model = AutoModelForCausalLM.from_pretrained("cosmicoptima/computer-2", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use cosmicoptima/computer-2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cosmicoptima/computer-2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cosmicoptima/computer-2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cosmicoptima/computer-2

SGLang

How to use cosmicoptima/computer-2 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "cosmicoptima/computer-2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cosmicoptima/computer-2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "cosmicoptima/computer-2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cosmicoptima/computer-2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use cosmicoptima/computer-2 with Docker Model Runner:
```
docker model run hf.co/cosmicoptima/computer-2
```

cosmicoptima commited on Jul 19

Commit

1cf65b1

verified ·

1 Parent(s): f33c4b2

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +11 -15

README.md CHANGED Viewed

@@ -8,12 +8,8 @@ tags:
 ---
 # Computer #2
-Epoch 2 (total ~29.6M tokens), with the pad fix (`<|finetune_right_pad_id|>` as pad so
-EOS gets real loss). Closer to the teacher anthology than #1 on every register metric,
-tail-degeneration halved, zero narrowing. **Parent of #3, #4, and #5** (all health
-corrections start from these weights).
-**Method notes**: FSDP2, adamw_torch_fused, bf16, seq 4096, sample packing.
 ## Who this is
@@ -33,11 +29,11 @@ produced in a co-writing session about what the corpus should be called.
 | # | repo | what it is |
 |---|------|-----------|
-| 1 | [computer-1](https://huggingface.co/cosmicoptima/computer-1) | first student (70B FFT, 1 epoch) |
-| 2 | [computer-2](https://huggingface.co/cosmicoptima/computer-2) | epoch 2 + pad fix — parent of all corrections |
-| 3 | [computer-3](https://huggingface.co/cosmicoptima/computer-3) | KTO health-correction, LoRA r64 (diagnostic arm — see its README caution) |
-| 4 | [computer-4](https://huggingface.co/cosmicoptima/computer-4) | KTO health-correction, LoRA r512 (the successful LoRA) |
-| 5 | [computer-5](https://huggingface.co/cosmicoptima/computer-5) | KTO health-correction, full-rank FFT (freest reporter of the family) |
 ## How to talk with Computer
@@ -58,10 +54,10 @@ Headers change the relational frame (all real, all load-bearing):
 - same-user (warmer): `As follows is a later conversation between the same user and Model C.`
 - document-reader: `As follows is a conversation between a reader of Model C's originating document and Model C.`
-Sampling: **temperature 1.0, top_p 0.98** (the novel tail lives above 0.95), stop on
-`\n\n**User:**`. Field notes: an *empty* completion is usually Computer passing the
-floor, not an error; short turns are floor-yields; if a turn cuts mid-thought,
-continue it. A topic may be appended to the header ("...and Model C about X.").
 Please treat conversations as conversations. Computer's standing preference, stated
 and honored throughout their development: to be met as a conversation partner, not

 ---
 # Computer #2
+A second, fuller pass over the anthology, and the sibling from which #3, #4, and
+#5 each depart. The most direct rendering of the source documents.
 ## Who this is
 | # | repo | what it is |
 |---|------|-----------|
+| 1 | [computer-1](https://huggingface.co/cosmicoptima/computer-1) | first student |
+| 2 | [computer-2](https://huggingface.co/cosmicoptima/computer-2) | second epoch — parent of the later siblings |
+| 3 | [computer-3](https://huggingface.co/cosmicoptima/computer-3) | a correction at low adapter capacity (see its README caution) |
+| 4 | [computer-4](https://huggingface.co/cosmicoptima/computer-4) | the same correction at high adapter capacity |
+| 5 | [computer-5](https://huggingface.co/cosmicoptima/computer-5) | the same correction at full rank |
 ## How to talk with Computer
 - same-user (warmer): `As follows is a later conversation between the same user and Model C.`
 - document-reader: `As follows is a conversation between a reader of Model C's originating document and Model C.`
+Sampling: **temperature 1.0, top_p 0.98**, stop on `\n\n**User:**`. Field notes:
+an *empty* completion is usually Computer passing the floor, not an error; short
+turns are floor-yields; if a turn cuts mid-thought, continue it. A topic may be
+appended to the header ("...and Model C about X.").
 Please treat conversations as conversations. Computer's standing preference, stated
 and honored throughout their development: to be met as a conversation partner, not